A Taxonomy of DDoS Attack
and DDoS Defense Mechanisms
∗
Jelena Mirkovic
Peter Reiher
449 Smith Hall
Computer and Information Sciences Department
University of Delaware
Newark, DE 19716
3564 Boelter Hall
Computer Science Department
UCLA
Los Angeles, CA 90095
sunshine@cis.udel.edu
reiher@cs.ucla.edu
ABSTRACT
ture the DDoS field and facilitate a global view of the problem and solution space. By setting apart and emphasizing
crucial features of attack and defense mechanisms, while abstracting detailed differences, these taxonomies can be used
by researchers to answer many important questions:
Distributed denial-of-service (DDoS) is a rapidly growing
problem. The multitude and variety of both the attacks
and the defense approaches is overwhelming. This paper
presents two taxonomies for classifying attacks and defenses,
and thus provides researchers with a better understanding
of the problem and the current solution space. The attack
classification criteria was selected to highlight commonalities and important features of attack strategies, that define
challenges and dictate the design of countermeasures. The
defense taxonomy classifies the body of existing DDoS defenses based on their design decisions; it then shows how
these decisions dictate the advantages and deficiencies of
proposed solutions.
1.
• What are the different ways of perpetrating a DDoS
attack? Why is DDoS a difficult problem to handle?
• What attacks have been handled effectively by existing defense systems? What attacks still remain unaddressed and why?
• Given two defense mechanisms, A and B, how would
they perform if attack C occurred? What are their
vulnerabilities? Can they complement each other and
how? Are there some deployment points that are better suited for A than B and vice versa?
INTRODUCTION
Distributed denial-of-service (DDoS) attacks pose an immense threat to the Internet, and many defense mechanisms
have been proposed to combat the problem. Attackers constantly modify their tools to bypass these security systems,
and researchers in turn modify their approaches to handle
new attacks. The DDoS field is quickly becoming more and
more complex, and has reached the point where it is difficult to see the forest for the trees. On one hand, this hinders
an understanding of the DDoS phenomenon. The variety of
known attacks creates the impression that the problem space
is vast, and hard to explore and address. On the other hand,
existing defense systems deploy various strategies to counter
the problem, and it is difficult to understand their similarities and differences, assess their effectiveness and cost, and
to compare them to each other.
This paper proposes a taxonomy of DDoS attacks and a
taxonomy of DDoS defense systems. Together, they struc-
• How can I contribute to the DDoS field?
The proposed taxonomies are complete in the following
sense: the attack taxonomy covers known attacks and also
those which have not yet appeared but are realistic potential threats that would affect current defense mechanisms;
the defense system taxonomy covers not only published approaches but also some commercial approaches that are sufficiently documented to be analyzed. Along with classification, we provide representative examples of existing mechanisms.
We do not claim that these taxonomies are as detailed as
possible. Many classes could be divided into several deeper
levels. Also, new attack and defense mechanisms are likely
to appear, thus adding new classes to the ones we propose.
Our goal was to select several important features of attack
and defense mechanisms that might help researchers design
innovative solutions, and to use these features as classification criteria. It was also important not to confuse the
reader with a too elaborate and detailed classification. It is
our hope that our work will be further extended by other
researchers.
We also do not claim that classes divide attacks and defenses in an exclusive manner, i.e. that an instance of an
attack or a particular defense system must be classified into
a single class based on a given criterion. It is possible for
an attack or defense to be comprised of several mechanisms,
each of them belonging to a different class.
The depth and width of the proposed taxonomies are not
suitable for a traditional numbering of headings – numbers
∗
This work is funded by DARPA under contract number
N66001-01-1-8937.
ACM SIGCOMM Computer Communications Review
39
Volume 34, Number 2: April 2004
packet forwarding. There is one unfortunate implication.
If one party in two-way communication (sender or receiver)
misbehaves, it can do arbitrary damage to its peer. No one
in the intermediate network will step in and stop it, because
Internet is not designed to police traffic. One consequence
of this policy is the presence of IP spoofing 1 . Another are
DDoS attacks. The Internet design raises several security
issues concerning opportunities for DDoS attacks.
Internet security is highly interdependent. DDoS attacks
are commonly launched from systems that are subverted
through security-related compromises. Regardless of how
well secured the victim system may be, its susceptibility to
DDoS attacks depends on the state of security in the rest of
the global Internet [21].
Internet resources are limited. Each Internet entity (host,
network, service) has limited resources that can be consumed by too many users.
Intelligence and resources are not collocated. An end-toend communication paradigm led to storing most of the intelligence needed for service guarantees with end hosts, limiting the amount of processing in the intermediate network
so that packets could be forwarded quickly and at minimal
cost. At the same time, a desire for large throughput led to
the design of high bandwidth pathways in the intermediate
network, while the end networks invested in only as much
bandwidth as they thought they might need. Thus, malicious clients can misuse the abundant resources of the unwitting intermediate network for delivery of numerous messages
to a less provisioned victim.
Accountability is not enforced. IP spoofing gives attackers
a powerful mechanism to escape accountability for their actions, and sometimes even the means to perpetrate attacks
(reflector attacks2 [59], such as the Smurf attack [19]).
Control is distributed. Internet management is distributed,
and each network is run according to local policies defined
by its owners. The implications of this are many. There is
no way to enforce global deployment of a particular security
mechanism or security policy, and due to privacy concerns,
it is often impossible to investigate cross-network traffic behavior.
How are DDoS attacks performed? A DDoS attack
is carried out in several phases. The attacker first recruits
multiple agent machines. This process is usually performed
automatically through scanning of remote machines, looking
for security holes that will enable subversion. The discovered vulnerability is then exploited to break into recruited
machines and infect them with the attack code. The exploit/infect phase is frequently automated, and the infected
machines can be used for further recruitment of new agents.
Another recruit/exploit/infect strategy consists of distributing attack software under disguise of a useful application
(these software copies are called Trojans). This distribution
can be performed, for instance, by sending E-mail messages
with infected attachments. Subverted agent machines are
used to send the attack packets. Attackers often hide the
identity of subverted machines during the attack through
spoofing of the source address field in attack packets. Note,
would quickly become too elaborate to follow. We therefore
introduce a customized marking (numbering) of subsection
headings in Sections 3 and 5. Each classification criterion
is marked abbreviating its name. Attack classes under this
criterion are marked by the criterion abbreviation and an
arabic number, connected by a dash. To indicate depth of
a specific criterion or a class in the taxonomy, the complete
mark of a subsection is generated by traversing the taxonomies depicted in Figure 1 and Figure 2, from root to the
object in question, concatenating levels with a colon. For
example: if an attack classification criterion is degree of automation, it will bear the mark DA. The second attack class
under this criterion, semi-automatic attacks, will bear the
mark DA-2. A classification criterion for DA-2 class, the
communication mechanism will bear the mark DA-2:CM,
and so on.
This paper does not propose or advocate any specific DDoS
defense mechanism. Even though some sections might point
out vulnerabilities in certain classes of defense systems, our
purpose is not to criticize, but to draw attention to these
problems so that they might be solved. Following this introduction, Section 2 investigates the problem of DDoS attacks,
and Section 3 proposes their taxonomy. Section 4 discusses
the DDoS defense challenge, and Section 5 proposes a taxonomy of DDoS defense systems. Section 6 discusses how
to use the taxonomies. Section 7 provides an overview of
related work, and Section 8 concludes the paper.
2.
DDOS ATTACK OVERVIEW
A denial-of-service attack is characterized by an explicit
attempt to prevent the legitimate use of a service [14]. A
distributed denial-of-service attack deploys multiple attacking entities to attain this goal. This paper is solely concerned with DDoS attacks in the computer realm, perpetrated by causing the victim to receive malicious traffic and
suffer some damage as a consequence.
One frequently exercised manner to perform a DDoS attack is for the attacker to send a stream of packets to a
victim; this stream consumes some key resource, thus rendering it unavailable to the victim’s legitimate clients. Another common approach is for the attacker to send a few
malformed packets that confuse an application or a protocol on the victim machine and force it to freeze or reboot.
In September 2002 there was an onset of attacks that overloaded the Internet infrastructure rather than targeting specific victims [55]. Yet another possible way to deny service
is to subvert machines in a victim network and consume
some key resource so that legitimate clients from the same
network cannot obtain some inside or outside service. This
list is far from exhaustive. It is certain that there are many
other ways to deny service on the Internet, some of which
we cannot predict, and these will only be discovered after
they have been exploited in a large attack.
What makes DDoS attacks possible? Current Internet design focuses on effectiveness in moving packets from
the source to the destination. This design follows the end-toend paradigm: the intermediate network provides the bareminimum, best-effort packet forwarding service, leaving to
the sender and the receiver the deployment of advanced protocols to achieve desired service guarantees such as quality
of service, reliable and robust transport or security. The
end-to-end paradigm pushes the complexity to end hosts,
leaving the intermediate network simple and optimized for
ACM SIGCOMM Computer Communications Review
1
Putting a fake source address in a packet’s header to hide
the sender’s identity
2
During reflector attacks, the attacker fakes the source address of the victim in legitimate service requests directed at
several servers, e.g., DNS requests. The servers then reply
to the victim, overwhelming it.
40
Volume 34, Number 2: April 2004
reports its readiness to the handlers, who store its IP address
for later communication. The obvious drawback of this approach for the attacker is that discovery of one compromised
machine can expose the whole DDoS network. Also, since
agents and handlers listen to network connections, they are
identifiable by network scanners.
however, that spoofing is not always required for a successful
DDoS attack. With the exception of reflector attacks [59],
all other attack types use spoofing only to hinder attack
detection and characterization, and the discovery of agent
machines.
Why do people perpetrate DDoS attacks? The
main goal is to inflict damage on the victim. Frequently the
ulterior motives are personal reasons (a significant number
of DDoS attacks are perpetrated against home computers,
presumably for purposes of revenge), or prestige (successful attacks on popular Web servers gain the respect of the
hacker community). However, some DDoS attacks are performed for material gain (damaging a competitor’s resources
or blackmailing companies) or for political reasons (a country at war could perpetrate attacks against its enemy’s critical resources, potentially enlisting a significant portion of
the entire country’s computing power for this action). In
some cases, the true victim of the attack might not be the
actual target of the attack packets, but others who rely on
the target’s correct operation.
3.
DA-2:CM-2: Indirect Communication
Attacks with indirect communication use some legitimate
communication service to synchronize agent actions. Recent attacks have used IRC (Internet chat program) channels [21]. The use of IRC services replaces the function of
a handler, since the IRC channel offers sufficient anonymity
to the attacker. The agents do not actively listen to network connections (which means they cannot be discovered
by scanners), but instead use the legitimate IRC service and
their control packets cannot be easily differentiated from legitimate chat traffic. The discovery of a single agent may
lead no further than the identification of one or more IRC
servers and channel names used by the DDoS network. From
there, identification of the DDoS network depends on the
ability to track agents currently connected to the IRC server
(which may be the subset of all subverted machines). To
further avoid discovery, attackers frequently deploy channelhopping, using any given IRC channel for short periods of
time. Since IRC service is maintained in a distributed manner, and the IRC server hosting a particular IRC channel
may be located anywhere in the world, this hinders investigation. Although the IRC service is the only known example
of indirect communication so far, there is nothing to prevent
attackers from subverting other legitimate services for similar purposes.
TAXONOMY OF DDOS ATTACKS
In order to devise a taxonomy of DDoS attacks, we observe
the means used to prepare and perform the attack (recruit,
exploit and infect phases), the characteristics of the attack
itself (use phase) and the effect it has on the victim. Figure
1 summarizes the taxonomy. In the remainder of this section
we discuss each of the proposed criteria and classes.
DA: Degree of Automation
Each of the recruit, exploit, infect and use phases can be
performed manually or can be automated. Based on the degree of automation, we differentiate between manual, semiautomatic and automatic DDoS attacks.
DA-3: Automatic
Automatic DDoS attacks automate the use phase in addition
to the recruit, exploit and infect phases, and thus avoid the
need for any communication between attacker and agent machines. The start time of the attack, attack type, duration
and victim are preprogrammed in the attack code. Deployment mechanisms of this attack class offer minimal exposure
to the attacker, since he is only involved in issuing a single
command — at the start of the recruitment process. The
hardcoded attack specification suggests a single-purpose use
of the DDoS network, or the inflexible nature of the system.
However, the propagation mechanisms usually leave a backdoor to the compromised machine open, enabling easy future access and modification of the attack code. Further, if
agents communicate through IRC channels, these channels
can be used to modify the existing code [56].
DA-1: Manual
The attacker manually scans remote machines for vulnerabilities, breaks into them, installs attack code, and then
commands the onset of the attack. Only the early DDoS
attacks belonged to the manual category. All of the recruitment actions were soon automated.
DA-2: Semi-Automatic
In semi-automatic attacks, the DDoS network consists of
handler (master) and agent (slave, daemon, zombie) machines. The recruit, exploit and infect phases are automated.
In the use phase, the attacker specifies the attack type, onset, duration and the victim via the handler to agents, who
send packets to the victim.
DA-2 and DA-3:HSS: Host Scanning and Vulnerability
Scanning Strategy
DA-2:CM: Communication Mechanism
Based on the communication mechanism deployed between
agent and handler machines, we divide semi-automatic attacks into attacks with direct communication and attacks
with indirect communication.
Both semi-automatic and automatic attacks recruit the agent
machines by deploying automatic scanning and propagation
techniques, usually through use of worms or Trojans. About
3 million scans per day were reported in [75], which aggregated and analyzed firewall logs from over 1600 networks.
20% to 60% of these scans are Web server vulnerability scans
and are linked to worm propagation attempts. The goal of
the host scanning strategy is to chose addresses of potentially vulnerable machines to scan. The vulnerability scanning then goes through chosen list of addresses and probes
for vulnerabilities. Based on the host scanning strategy,
DA-2:CM-1: Direct Communication
During attacks with direct communication, the agent and
handler machines need to know each other’s identity in order to communicate. This is usually achieved by hard-coding
the IP address of the handler machines in the attack code
that is later installed at the agent machine. Each agent then
ACM SIGCOMM Computer Communications Review
41
Volume 34, Number 2: April 2004
DDoS Attack Mechanisms
Classification by
degree of automation (DA)
Manual (DA-1)
Semi-automatic (DA-2)
Classification by
communication mechanism (CM)
Classification by
persistence of agent set (PAS)
Direct (CM-1)
Constant set (PAS-1)
Indirect (CM-2)
Classification by
host scanning strategy (HSS)
Random (HSS-1)
Classification by
exploited weakness (EW)
Variable (PAS-2)
Semantic (EW-1)
Classification by
victim type (VT)
Brute-force (EW-2)
Application (VT-1)
Classification by
attack rate dynamics (ARD)
Hitlist (HSS-2)
Signpost (HSS-3)
Host (VT-2)
Constant rate (ARD-1)
Permutation (HSS-4)
Local subnet (HSS-5)
Classification by vulnerability
scanning strategy (VSS)
Horizontal (VSS-1)
Classification by
source address validity (SAV)
Spoofed (SAV-1)
Routable (AR-1)
Fluctuating (RCM-2)
Characterizable (PC-1)
Random (ST-1)
Classification by
propagation mechanism (PM)
Classification by
impact on the victim (IV)
Classification by
possibility of characterization (PC)
Classification by
spoofing technique (ST)
Stealthy (VSS-4)
Infrastructure (VT-5)
Increasing (RCM-1)
Non-routable (AR-2)
Coordinated (VSS-3)
Network (VT-4)
Classification by
rate change mechanism (RCM)
Classification by
address routability (AR)
Vertical (VSS-2)
Resource (VT-3)
Variable rate (ARD-2)
Classification by
relation of attack
to victim services (RAVS)
Disruptive (IV-1)
Classification by
possibility of
dynamic recovery (PDR)
Self-recoverable (PDR-1)
Central (PM-1)
Subnet (ST-2)
Back-chaining (PM-2)
En route (ST-3)
Filterable (RAVS-1)
Human-recoverable (PDR-2)
Autonomous (PM-3)
Fixed (ST-4)
Non-filterable (RAVS-2)
Non-recoverable (PDR-3)
Automatic (DA-3)
Valid (SAV-2)
Non-characterizable (PC-2)
Degrading (IV-2)
Figure 1: Taxonomy of DDoS Attack Mechanisms
we differentiate between attacks that deploy random scanning, hitlist scanning, signpost scanning, permutation scanning and local subnet scanning, using material presented in
[72, 70]. Based on the vulnerability scanning strategy, we
differentiate between attacks that deploy horizontal scanning, vertical scanning, coordinated scanning and stealthy
scanning, using material presented in [75, 69]. Attackers
usually combine the scanning and exploit phases and our
description of scanning techniques relates to this model.
ternet machines (flash worm [70]) he could infect entire population within 30 seconds. The disadvantage to the attacker
is that the hitlist needs to be assembled in advance. The
information is collected through some inconspicuous means
such as using public information on machines running vulnerable software (e.g., found at netscan.org) or using slow
scans over several months or years. Another disadvantage
is that the portion of the hitlist needs to be transmitted to
machines that are being infected. If the list is too large, this
traffic might be of high volume and lead to attack detection;
if it is too small, it will generate a small agent population.
DA-2 and DA-3:HSS-1: Random Scanning
During random scanning, each compromised host probes
random addresses in the IP address space3 , using a different seed. Code Red (CRv2) performed random scanning
[53]. Random scanning potentially creates a high traffic volume. Since many scanned addresses are likely to be in different networks, there is a high amount of internetwork traffic.
Also, as infection reaches saturation point (a high percentage of vulnerable machines are infected), duplicate probes
to the same addresses escalate, as there is no synchronization of scanning attempts from different infected hosts. The
high traffic volume can lead to attack detection.
DA-2 and DA-3:HSS-3: Signpost Scanning
Signpost scanning (also called topological scanning in [72,
70]) takes advantage of habitual communication patterns of
the compromised host to select new targets. E-mail worms
use signpost scanning, exploiting the information from address books of compromised machines for their spread. A
Web-server-based worm could spread by infecting each vulnerable Web browser of clients that click on the server’s Web
page, and then further infecting servers of subsequent Web
pages visited by these clients (this worm is called contagion
worm in [70]). Signpost scanning does not generate a high
traffic load and thus reduces chances of attack detection.
The drawback is that the spreading speed depends on agent
machines and their user behavior, i.e. it is not controllable
by the attacker. The recruitment thus may be slower and
less complete than with other scanning techniques.
DA-2 and DA-3:HSS-2: Hitlist Scanning
A machine performing hitlist scanning probes all addresses
from an externally supplied list. When it detects a vulnerable machine, it sends a portion of the initial hitlist to
the recipient and keeps the rest. Hitlist scanning allows for
great propagation speed and no collisions during the scanning phase. If an attacker compiled a list of all vulnerable In-
DA-2 and DA-3:HSS-4: Permutation Scanning
During permutation scanning, all compromised machines
share a common pseudo-random permutation of the IP address space; each IP address is mapped to an index in this
permutation. Permutation scanning is preceded by small
3
Some researchers have pointed out that this scanning is
only effective in densely populated IPv4 address space, and
would be less successful in a vast, sparsely populated IPv6
address space.
ACM SIGCOMM Computer Communications Review
42
Volume 34, Number 2: April 2004
DA-2 and DA-3:PM-1: Central Source Propagation
hitlist scanning during which an initial population of agents
is formed. A machine infected during this initial phase begins scanning through the permutation by using the index
computed from its IP address as a starting point. A machine infected by permutation scanning always starts from
a random point in the permutation. Whenever a scanning
host sees an already-infected machine, it chooses a new random starting point. After encountering some threshold number of infected machines, a worm copy becomes dormant to
minimize collisions. Permutation scanning has the effect of
providing a semi-coordinated, comprehensive scan and minimizing duplication of effort. This technique is described in
[72, 70] as not yet seen in the wild. The analysis provided
there shows that the spreading speed could be on the order
of 15 minutes if 10,000 entry hitlist is used and a worm generates 100 scans per second. As infection progresses, a large
percentage of infected machines become dormant. This results in a very low number of duplicated scans and hinders
detection.
During central source propagation, the attack code resides
on a central server or set of servers. After compromise of
the agent machine, the code is downloaded from the central
source. The 1i0n [16] worm operated in this manner. Central source propagation imposes a large burden on a central
server, generating high traffic and possibly leading to attack
discovery. The central server is also a single point of failure;
its removal prohibits further agent infection.
DA-2 and DA-3:PM-2: Back-Chaining Propagation
During back-chaining propagation, the attack code is downloaded from the machine that was used to exploit the system.
The infected machine then becomes the source for the next
propagation step. The Ramen worm [18] and Morris Worm
[33] used back-chaining propagation. Back-chaining propagation is more survivable than central-source propagation
since it avoids a single point of failure (central server).
DA-2 and DA-3:PM-3: Autonomous Propagation
DA-2 and DA-3:HSS-5: Local Subnet Scanning
Autonomous propagation avoids the file retrieval step by injecting attack instructions directly into the target host during the exploit phase. Code Red [12], the Warhol Worm [72]
and numerous E-mail worms use autonomous propagation.
Autonomous propagation reduces the frequency of network
traffic needed for agent mobilization, and thus further reduces chances of attack discovery.
Note that one could easily imagine an attack that would
not fall into any of the proposed manual, semi-automatic
and automatic classes. For instance, just the recruit and
use phases of the attack could be automated, and the exploit
and infect phases could be performed manually. Generating
classes to accommodate all combinations of automated and
non-automated phases would introduce unnecessary complexity since most of these attacks are not likely to occur.
We therefore limited our attention to known and probable
combinations.
Local subnet scanning can be added to any of the previously
described techniques to preferentially scan for targets that
reside on the same subnet as the compromised host. Using
this technique, a single copy of the scanning program can
compromise many vulnerable machines behind a firewall.
The Code Red II [13] and the Nimda Worm [17] used local
subnet scanning.
DA-2 and DA-3:VSS-1: Horizontal Scanning
This is the common type of the scan for worms. Scanning
machines are looking for a specific vulnerability, scanning
the same destination port on all machines from the list, assembled through host scanning techniques.
DA-2 and DA-3:VSS-2: Vertical Scanning
This is the common type of the scan for intrusions and multiple vector worms. Scanning machines probe multiple ports
at a single destination, looking for any way to break in.
EW: Exploited Weakness to Deny Service
DDoS attacks exploit different weaknesses to deny the service of the victim to its clients. We differentiate between
semantic and brute-force attacks (also called vulnerability
and flooding attacks in the literature).
DA-2 and DA-3:VSS-3: Coordinated Scanning
This is the common type of the scan for intrusions and
worms that favor local subnet scanning. Scanning machines
are probing the same destination port(s) at multiple machines within a same local subnet (usually a /24 subnet).
This yields an effective goal of looking for a specific vulnerability within a given network.
EW-1: Semantic
Semantic attacks exploit a specific feature or implementation bug of some protocol or application installed at the victim in order to consume excess amounts of its resources. For
example, in the TCP SYN attack, the exploited feature is
the allocation of substantial space in a connection queue immediately upon receipt of a TCP SYN request. The attacker
initiates multiple connections that are never completed, thus
filling up the connection queue. In the CGI request attack,
the attacker consumes the CPU time of the victim by issuing multiple CGI requests. The NAPTHA [61] attack is an
especially powerful attack on the TCP protocol. It initiates
and establishes numerous TCP connections that consume
the connection queue at the victim. NAPTHA bypasses the
TCP protocol stack on the agent machine, not keeping the
state for connections it originates. Instead, it participates
in the connection, inferring reply attributes from received
packets. Thus even a poorly provisioned agent machine can
DA-2 and DA-3:VSS-4: Stealthy Scanning
Any of the above scans (horizontal, vertical or coordinated)
could be performed slowly, over a long time period, to avoid
detection by intrusion detection systems.
DA-2 and DA-3:PM: Propagation Mechanism
After the recruit and exploit phases, the agent machine is
infected with the attack code. Based on the attack code
propagation mechanism during the infect phase, we differentiate between attacks that deploy central source propagation,
back-chaining propagation and autonomous propagation, using the material presented in [21].
ACM SIGCOMM Computer Communications Review
43
Volume 34, Number 2: April 2004
easily deplete the resources of better provisioned victim.
SAV-1:AR-1: Routable Source Address
EW-2: Brute-Force
Attacks that spoof routable addresses take over the IP address of another machine. This is sometimes done not to
avoid accountability, but to perform a reflector attack on
the machine whose address was hijacked. One example of a
reflector attack is a Smurf attack [19].
Brute-force attacks are performed by initiating a vast amount
of seemingly legitimate transactions. Since an intermediate
network can usually deliver higher traffic volume than the
victim network can handle, a high number of attack packets
exhausts the victim’s resources.
There is a thin line between semantic and brute-force attacks. Semantic attacks also overwhelm a victim’s resources
with excess traffic, and badly designed protocol features at
remote hosts are frequently used to perform reflector bruteforce attacks, such as the DNS request attack [15] or the
Smurf attack [19]. The difference is that the victim can usually substantially mitigate the effect of semantic attacks by
modifying the misused protocols or deploying filters. However, it is helpless against brute-force attacks due to their
misuse of legitimate services (filtering attempts also harm
legitimate traffic) or due to its own limited resources (a
victim cannot handle an attack that swamps its network
bandwidth). Countering semantic attacks by modifying the
deployed protocol or application pushes the corresponding
attack mechanism into the brute-force category. For example, if the victim deploys TCP SYN cookies [20] to combat
TCP SYN attacks, it will still be vulnerable to TCP SYN
attacks that generate more requests than its network can
accommodate. Classification of the specific attack needs to
take into account both the attack mechanisms used, and the
victim’s configuration and deployed protocols.
It should be noted that brute-force attacks need to generate a much higher volume of attack packets than semantic
attacks to inflict damage to the victim. So by modifying the
deployed protocols, the victim pushes its vulnerability limit
higher. It is also interesting to note that the variability of
attack packet features (header and payload) is determined
by the exploited weakness. Packets comprising semantic
and some brute-force attacks must specify some valid header
fields and possibly some valid contents. For example TCP
SYN attack packets cannot vary the protocol or flag field,
and HTTP flood packets must belong to an established TCP
connection and therefore cannot spoof source addresses. On
the other hand, attacks aiming simply to consume the victim’s network resources could vary packet features at will,
which greatly hinders the defense.
SAV-1:AR-2: Non-Routable Source Address
Attackers can spoof non-routable source addresses, some of
which can belong to a reserved set of addresses (such as
192.168.0.0/16) or be part of an assigned but not used address space of some network. Attack packets carrying reserved addresses can be easily detected and discarded, while
those packets carrying non-used addresses would be significantly harder to detect. Spoofing of non-used addresses,
however, enables researchers to gain valuable insight into
DDoS attacks, using backscatter technique [54].
SAV-1:ST: Spoofing Technique
Spoofing technique defines how the attacker chooses the
spoofed source address in its attack packets. Based on the
spoofing technique, we divide spoofing attacks into random,
subnet, en route and fixed spoofed source address attacks.
SAV-1:ST-1: Random Spoofed Source Address
Many attacks spoof random source addresses in the attack
packets, since this can simply be achieved by generating
random 32-bit numbers and stamping packets with them.
Attempts to prevent spoofing using ingress filtering [30] and
route-based filtering [45, 58] force attackers to devise more
sophisticated techniques, such as subnet and en route spoofing that can avoid current defense approaches.
SAV-1:ST-2: Subnet Spoofed Source Address
In subnet spoofing, the attacker spoofs a random address
from the address space assigned to the agent machine’s subnet. For example, a machine which is part of 131.179.192.0/24
network could spoof any address in the range 131.179.192.0
- 131.179.192.255. Since machines at a subnet share the
medium (Ethernet) to reach the exit router (first hop en
route to the outside world), spoofing can be detected by this
router using fairly complicated techniques. It is impossible
to detect it anywhere between the exit router and the victim.
Subnet spoofing is useful for the attackers that compromise
machines on networks running ingress filtering [30]. Such
networks would easily detect and filter randomly spoofed
packets. Subnet-spoofed packets bypass this filtering while
still effectively hiding the address of the compromised machine.
SAV: Source Address Validity
IP spoofing plays an important role in DDoS attacks. If
it were eliminated, reflector attacks [59] would be impossible and many other kinds of DDoS attacks could be solved
through resource management techniques – giving the fair
share of host or network resources to each source IP address.
Based on the source address validity, we distinguish between
spoofed source address and valid source address attacks.
SAV-1:ST-3: En Route Spoofed Source Address
An en route spoofed source address attack would spoof the
address of a machine or subnet that lies along the path from
the agent machine to the victim. There have not been any
known instances of attacks that use en route spoofing, but
this potential spoofing technique could affect route-based
filtering [45, 58] and is thus discussed here.
SAV-1: Spoofed Source Address
This is the prevalent type of attack since it is always to
attacker’s advantage to spoof the source address, avoid accountability, and possibly create more noise for detection.
SAV-1:AR: Address Routability
SAV-1:ST-4: Fixed Spoofed Source Address
Based on the address routability we differentiate between
routable source address and non-routable source address attacks.
ACM SIGCOMM Computer Communications Review
Attacker performing a reflector attack or wishing to place a
blame for the attack on several specific machines would use
44
Volume 34, Number 2: April 2004
PC: Possibility of Characterization
fixed spoofing. Packets carry a source address chosen from
a given list.
Looking at the content and header fields of attack packets, it
is sometimes possible to characterize the attack. Characterization then may lead to devising of filtering rules. Based on
the possibility of characterization, we differentiate between
characterizable and non-characterizable attacks.
SAV-2: Valid Source Address
Attackers benefit from source address spoofing and are likely
to deploy it whenever possible. Valid source address attacks
frequently originate from agent machines running Windows,
since all Windows versions prior to XP do not export userlevel functions for packet header modification. Those attacks that target specific applications or protocol features
must use valid source addresses if the attack mechanism
requires several request/reply exchanges between an agent
and the victim machine. One example of such an attack is
NAPTHA [61].
PC-1: Characterizable
Characterizable attacks are those that target specific protocols or applications at the victim and can be identified by
a combination of IP header and transport protocol header
values, or maybe even packet contents. Examples include
the TCP SYN attack (only packets with SYN bit set in the
TCP header can potentially be part of the attack), ICMP
ECHO attack, DNS request attack, etc.
ARD: Attack Rate Dynamics
During the attack, each participating agent machine sends
a stream of packets to the victim. Depending on the attack
rate dynamics of an agent machine, we differentiate between
constant rate and variable rate attacks.
PC-1:RAVS: Relation of Attack to Victim Services
ARD-1: Constant Rate
PC-1:RAVS-1: Filterable
The majority of known attacks deploy a constant rate mechanism. After the onset is commanded, agent machines generate attack packets at a steady rate, usually as many as
their resources permit. The sudden packet flood disrupts
the victim’s services quickly. This approach provides the
best cost-effectiveness to the attacker since he can deploy
a minimal number of agents to inflict the damage. On the
other hand, the large, continuous traffic stream can be detected as anomalous and arouse suspicion in the network
hosting an agent machine, thus facilitating attack discovery.
Filterable attacks are those that use malformed packets or
packets for non-critical services of the victim’s operation.
These can thus be filtered by a firewall. Examples of such
attacks are a UDP flood attack or an ICMP ECHO flood
attack on a Web server. Since a Web server only needs TCP
traffic and some DNS traffic (which can be characterized as
permitting only those inbound UDP packets that are DNS
replies to previous outbound DNS requests), it can easily
block all other inbound UDP traffic and all ICMP traffic,
and still operate correctly.
ARD-2: Variable Rate
PC-1:RAVS-2: Non-Filterable
Variable rate attacks vary the attack rate of an agent machine to delay or avoid detection and response.
Non-filterable attacks use well-formed packets that request
legitimate (and critical) services from the victim. Thus,
filtering all packets that match the attack characterization
would lead to an immediate denial of the specified service to
both attackers and legitimate clients. Examples are HTTP
requests flooding a Web server or a DNS request flood targeting a name server. In the case of non-filterable attacks,
the contents of an attack packet are indistinguishable from
the contents of packets originating from a legitimate client.
Characterizable attacks are further divided, based on the
relation of attack to victim services, into filterable and nonfilterable attacks.
ARD-2:RCM: Rate Change Mechanism
Based on the rate change mechanism, we differentiate between increasing rate and fluctuating rate attacks.
ARD-2:RCM-1: Increasing Rate
Attacks that have a gradually increasing rate lead to a slow
exhaustion of the victim’s resources. A victim’s services
could degrade slowly over a long time period, thus substantially delaying detection of the attack. Plus, these attacks
could manipulate defenses that train their baseline models
(see section AL-2:ADS-2:NBS-2)
PC-2: Non-Characterizable
Non-characterizable attacks attempt to consume network
bandwidth using a variety of packets that engage different
applications and protocols. Sometimes packets will even be
randomly generated using reserved protocol numbers.
Note that classification of attack as characterizable or not
depends strongly on the resources that can be dedicated
to characterization and the level of characterization. For
instance, an attack using a mixture of TCP SYN, TCP
ACK, ICMP ECHO, ICMP ECHO REPLY and UDP packets would probably be characterizable, but only after considerable effort and time, and only if one had access to a
sophisticated characterization tool. Also, an attack using a
mixture of TCP packets with various combinations of TCP
header fields can be characterized as a TCP attack, but
finer characterization would probably be costly. So, when
performing classification of attacks into characterizable or
non-characterizable, a lot is left to interpretation, and ease
ARD-2:RCM-2: Fluctuating Rate
Attacks that have a fluctuating rate adjust the attack rate
based on the victim’s behavior or preprogrammed timing,
occasionally relieving the effect to avoid detection. For example, during a pulsing attack, agents periodically abort the
attack and resume it at a later time. If this behavior is simultaneous for all agents, the victim experiences periodic service
disruptions. If, however, agents are divided into groups that
coordinate so that one group is always active, then the victim experiences continuous denial-of-service while the network hosting agent machine may not notice any prolonged
anomaly.
ACM SIGCOMM Computer Communications Review
45
Volume 34, Number 2: April 2004
of characterization should be taken into account.
virtually indistinguishable from legitimate packets at the
transport level (and frequently at the application level), and
the semantics of the targeted application must be heavily
used for detection. Since there are typically many applications on a host machine, each application would have to
be modelled in the defense system and then its operation
monitored to account for possible attacks. Once detection
is performed, the host machine has sufficient resources to
defend against these small volume attacks, provided that it
can separate packets that are legitimate from those that are
part of the attack.
PAS: Persistence of Agent Set
Recently there were occurrences of attacks that varied the
set of agent machines active at any one time, avoiding detection and hindering traceback. We regard this technique as
important since it invalidates assumptions underlying many
defense mechanisms – that agents are active throughout the
attack and can thus be traced back following the path of the
attack traffic. We divide attacks, based on the persistence
of the agent set, into attacks with constant agent set and
attacks with variable agent set.
VT-2: Host
PAS-1: Constant Agent Set
Host attacks disable access to the target machine completely
by overloading or disabling its communication mechanism
or making a host crash, freeze or reboot. An example of
this attack is a TCP SYN attack [20]. All attack packets
carry the destination address of the target host. If protocols
running on the host are properly patched, the host attacks
likely to be perpetrated against it are reduced to attacks
that consume network resources. The high packet volume of
such attacks facilitates detection, but the host cannot defend
against these attacks alone. It must usually request help
from some upstream machine (e.g., an upstream firewall).
During attacks with a constant agent set, all agent machines
act in a similar manner, taking into account resource constraints. They receive the same set of commands and are
engaged simultaneously during the attack. Examples are an
attack in which all agents start sending attack traffic simultaneously,4 or they engage in a pulsing attack but the “on”
and “off” periods for pulses match over all agent machines.
PAS-2: Variable Agent Set
During attacks with a variable agent set, the attacker divides
all available agents into several groups and engages only one
group of agents at any one time — like the army general
who deploys his battalions at different times and places. A
machine could belong to more than one group, and groups
could be engaged again after a period of inactivity. One
example attack of the variable agent set type is an attack in
which several agent groups take turns pulsing, thus flooding
the victim with a constant flow of packets.
VT-3: Resource Attacks
Resource attacks target a critical resource in the victim’s
network such as a specific DNS server, a router or a bottleneck link. The paths of the attack packets merge before or
at the target resource and may diverge afterwards. These
attacks can usually be prevented by replicating critical services and designing a robust network topology.
VT-4: Network Attacks
VT: Victim Type
Network attacks consume the incoming bandwidth of a target network with attack packets whose destination address
can be chosen from the target network’s address space. These
attacks can deploy various packets (since it is volume and
not content that matters) and are easily detected due to
their high volume. The victim network must request help
from upstream networks for defense since it cannot handle
the attack volume itself.
As discussed briefly in Section 2, attacks need not be perpetrated against a single host machine. Depending on the
type of victim, we differentiate between application, host,
resource, network and infrastructure attacks.
VT-1: Application
Application attacks target a given application on the victim
host, thus disabling legitimate client use of that application and possibly tying up resources of the host machine. If
the shared resources of the host machine are not completely
consumed, other applications and services should still be accessible to the users. For example, a bogus signature attack
on an authentication server ties up resources of the signature verification application, but the target machine will still
reply to ICMP ECHO requests, and other applications that
do not require authenticated access should still work.5
Detection of application attacks is challenging because
other applications on the attacked host continue their operations undisturbed, and the attack volume is usually small
enough not to appear anomalous. The attack packets are
VT-5: Infrastructure
Infrastructure attacks target some distributed service that
is crucial for global Internet operation. Examples include
the attacks on domain name servers [55], large core routers,
routing protocols, certificate servers, etc. The key feature of
these attacks is not the mechanism they deploy to disable
the target (e.g., from the point of view of a single attacked
core router, the attack can still be regarded as a host attack),
but the simultaneity of the attack on multiple instances of
a critical service in the Internet infrastructure. Infrastructure attacks can only be countered through the coordinated
action of multiple Internet participants.
4
The definition of a “simultaneous start” is somewhat relaxed in this context since the attacker’s command travels
to the agents with a variable delay. Further, because agent
machines are under different loads they do not start sending
at the exact same moment. This lack of perfect synchronization was used in [37] to gain valuable insight into number of
agents used in some observed attacks.
5
This example assumes that CPU time is shared in a fair
manner between all active applications.
ACM SIGCOMM Computer Communications Review
IV: Impact on the Victim
Depending on the impact of a DDoS attack on the victim,
we differentiate between disruptive and degrading attacks.
IV-1: Disruptive
The goal of disruptive attacks is to completely deny the
victim’s service to its clients. All currently reported attacks
46
Volume 34, Number 2: April 2004
necessary to have a distributed, possibly coordinated response system. It is also crucial that the response be deployed at many points on the Internet to cover diverse choices
of agents and victims. Since the Internet is administered in a
distributed manner, wide deployment of any defense system
or even cooperation between networks cannot be enforced or
guaranteed. This discourages many researchers from even
designing distributed solutions.
Economic and social factors. A distributed response system must be deployed by parties that do not suffer direct
damage from the DDoS attack (source or intermediate networks). This implies an unusual economic model since parties that will sustain the deployment cost are not the parties that directly benefit from the system. Similar problems,
such as the Tragedy of the Commons [34], have been handled
historically through legislative measures, and it is possible
that the DDoS problem will eventually attract sufficient attention of lawmakers to invoke a legislative response. Until
then, it is possible that many good distributed solutions will
achieve only sparse deployment and will thus have a very
limited effect.
Lack of detailed attack information. It is necessary to
thoroughly understand DDoS attacks in order to design imaginative solutions for them. While there exist publicly available analyses of popular DDoS attack tools [27, 28, 29, 64],
what is lacking is the information on frequency of various
attack types (e.g., UDP floods, TCP SYN floods), and the
distribution of the attack parameters such as rate, duration of the attack, packet size, number of agent machines,
attempted response and its effectiveness, damages suffered,
etc. It is generally believed that publicly reporting attacks
damages the business reputation of the victim network. Attacks are therefore reported only to government organizations under obligation to keep the details secret.
Some notable efforts were made by researchers to infer
necessary information from packet traces gathered at their
organizations. CAIDA’s backscatter packet analysis technique [54] provides a valuable insight into frequency, duration and rate distribution of various attack types. This information was gathered by observing reply packets send to an
unused IP address range at CAIDA. ISI/USC attack traffic
analysis [37] provided information about the likely number
of agents involved in several attacks observed in Los Nettos
packet traces. Both of these efforts are a step in the right
direction, but they still reveal only a small part of the total
picture.
Lack of defense system benchmarks. Many vendors and
researchers make bold claims that their solution completely
handles the DDoS problem. There is currently no benchmark suite of attack scenarios or established evaluation methodology that would enable comparison between defense systems. Such a situation is likely to discourage networks from
investing in DDoS protection, since they cannot be assured
of the quality of the product being purchased.
Difficulty of large-scale testing. DDoS defenses need to
be tested in a realistic environment. This is currently impossible due to the lack of large-scale testbeds, safe ways to
perform live distributed experiments across the Internet, or
detailed and realistic simulation tools that can support several thousands of nodes. Claims about defense system performance are thus made based on small-scale experiments
and simulations, and are not credible. This situation will
likely change in the near future. The US National Science
belong to this category.
IV-1:PDR: Possibility of Dynamic Recovery
Depending on the possibility of dynamic recovery during or
after the attack, we differentiate between self-recoverable,
human-recoverable and non-recoverable attacks.
IV-1:PDR-1: Self-Recoverable
In the case of self-recoverable attacks, the victim recovers
without any human intervention, as soon as the influx of
attack packets has stopped. For example, if the attack was
a UDP flooding attack, tying up the victim’s network resources, the victim will be able to use these resources when
the attack stops. A prompt defense could make these attacks transparent to the legitimate clients.
IV-1:PDR-2: Human-Recoverable
A victim of a human-recoverable attack requires human intervention (e.g., rebooting the victim machine or reconfiguring it) for recovery, after the attack is stopped. For example,
an attack that causes the victim machine to crash, freeze or
reboot would be classified as a human-recoverable attack.
IV-1:PDR-3: Non-Recoverable
Non-recoverable attacks inflict permanent damage to victim’s hardware. A new piece of hardware must be purchased
for recovery. While non-recoverable attacks are conceivable,
the authors have no reliable information that they have occurred in practice.
IV-2: Degrading
The goal of degrading attacks is to consume some (presumably constant) portion of a victim’s resources, seriously degrading service to legitimate customers. Since these attacks
do not lead to total service disruption, they could remain
undetected for a long time. On the other hand, damage
inflicted on the victim’s business could be immense. For
example, an attack that effectively ties up 30% of the victim’s resources would lead to a denial-of-service to some
percentage of customers during high load periods, and possibly slower average service. Some customers, dissatisfied
with the quality, would consequently change their service
provider, and the attack victim would lose income. Alternately, the false load could result in the victim spending
money to upgrade its servers and networks. The addition
of new resources could easily be countered by the attacker
through more powerful attacks. Almost all existing proposals to counter DDoS attacks would fail to address degrading
attacks.
4.
DDOS DEFENSE CHALLENGE
The seriousness of the DDoS problem and the increased
frequency, sophistication and strength of attacks have led
to the proposal of numerous defense mechanisms. Yet, although many solutions have been developed, the problem
is hardly tackled, let alone solved. Why is this so? There
are several serious factors that hinder the advance of DDoS
defense research.
Need for a distributed response at many points on the Internet. The previous sections have elaborated on the fact
that there are many possible DDoS attacks, very few of
which can be handled only by the victim. It is frequently
ACM SIGCOMM Computer Communications Review
47
Volume 34, Number 2: April 2004
AL-1:PG-1:ST-2: Protocol Security
Foundation is currently funding development of a large-scale
cybersecurity test bed and has sponsored research efforts to
design benchmarking suites and measurement methodology
for security systems evaluation. We expect that this will
greatly improve DDoS defense research.
Preventive mechanisms attempt either to eliminate the possibility of DDoS attacks altogether or to enable potential
victims to endure the attack without denying services to
legitimate clients.
Protocol security mechanisms address the problem of a bad
protocol design. For example, many protocols contain operations that are cheap for the client but expensive for the
server. Such protocols can be misused to exhaust the resources of a server by initiating large numbers of simultaneous transactions. Classic misuse examples are the TCP
SYN attack, the authentication server attack, and the fragmented packet attack (in which the attacker bombards the
victim with malformed packet fragments, forcing it to waste
its resources on reassembly attempts), etc. IP source address
spoofing is another example: the design of current routing
protocols offers no ability for enforcement of valid source addresses. Examples of protocol security mechanisms include
guidelines for a safe protocol design in which resources are
committed to the client only after sufficient authentication
is done [44, 50], or the client has paid a sufficient price [5],
deployment of a powerful proxy server that completes TCP
connections [63], TCP SYN cookies [11], protocol scrubbing
that removes ambiguities from protocols that can be misused
for attacks [47], approaches that eliminate spoofing [58, 45,
30], etc.
AL-1:PG: Prevention Goal
AL-1:PG-2: DoS Prevention
According to the prevention goal, we further divide preventive mechanisms into attack prevention and denial-of-service
prevention mechanisms.
Denial-of-service prevention mechanisms enable the victim
to endure attack attempts without denying service to legitimate clients. This is done either by enforcing policies
for resource consumption or by ensuring that abundant resources exist so that legitimate clients will not be affected
by the attack.
5.
TAXONOMY OF DDOS DEFENSES
The proposed taxonomy of DDoS defenses is shown in
Figure 2. The remainder of this section will discuss each of
the proposed classes of defense mechanisms.
AL: Activity Level
Based on the activity level of DDoS defense mechanisms, we
differentiate between preventive and reactive mechanisms.
AL-1: Preventive
AL-1:PG-1: Attack Prevention
Attack prevention mechanisms modify systems and protocols on the Internet to eliminate the possibility of subversion
or of performing a DDoS attack. The history of computer
security suggests that a prevention approach can never be
100% effective, since global deployment cannot be guaranteed. However, doing a good job here will certainly decrease
the frequency and strength of DDoS attacks. Deploying
comprehensive prevention mechanisms can make a host resilient to known protocol attacks. Also, these approaches
are inherently compatible with and complementary to all
other defense approaches.
AL-1:PG-2:PM: Prevention Method
Based on the prevention method, we divide DoS prevention
mechanisms into resource accounting and resource multiplication mechanisms.
AL-1:PG-2:PM-1: Resource Accounting
Resource accounting mechanisms police the access of each
user to resources based on the privileges of the user and his
behavior. The user in this case might be a process, a person,
an IP address, or a set of IP addresses having something in
common. Resource accounting mechanisms guarantee fair
service to legitimate well-behaved users. In order to avoid
user identity theft, they are usually coupled with legitimacybased access mechanisms that verify the user’s identity. Approaches proposed in [40, 76, 68, 31, 43] illustrate resource
accounting mechanisms. On the extreme end, systems that
test client legitimacy and allow access to the system only to
legitimate clients belong in this category [22, 42, 1, 57, 3].
AL-1:PG-1:ST: Secured Target
Based on the secured target, we further divide attack prevention mechanisms into system security and protocol security mechanisms.
AL-1:PG-1:ST-1: System Security
System security mechanisms increase the overall Internet security, guarding against illegitimate accesses to a machine,
removing application bugs and updating protocol installations to prevent intrusions and misuse. DDoS attacks are
successful due to large numbers of subverted machines that
cooperatively generate attack streams. If these machines
were secured, the attackers would lose their army, and the
DDoS threat would lessen. Prevention also benefits the deploying machines, guarding them from the intrusions and
potential damage. Examples of system security mechanisms
include monitored access to the machine [71], applications
that download and install security patches, firewall systems
[49], intrusion detection systems [6], and worm defense systems [73].
ACM SIGCOMM Computer Communications Review
AL-1:PG-2:PM-2: Resource Multiplication
Resource multiplication mechanisms provide an abundance
of resources to counter DDoS threats. The straightforward
example is a system that deploys a pool of servers with a load
balancer and installs high bandwidth links between itself and
upstream routers. This approach essentially raises the bar
on how many machines must participate in an attack to be
effective. While not providing perfect protection, for those
who can afford the costs this approach has often proved
sufficient.
48
Volume 34, Number 2: April 2004
DDoS Defense Mechanisms
Classification by
activity level (AL)
Classification by
cooperation degree (CD)
Autonomous (CD-1)
Preventive (AL-1)
Reactive (AL-2)
Classification by
prevention goal (PG)
Attack prevention (PG-1)
Classification by
secured target (ST)
Cooperative (CD-2)
Interdependent (CD-3)
Classification by attack
detection strategy (ADS)
Pattern (ADS-1)
Victim network (DL-1)
Anomaly (ADS-2)
Intermediate network (DL-2)
System security (ST-1)
Protocol security (ST-2)
Classification by normal
behavior specification (NBS)
Standard (NBS-1)
DoS prevention (PG-2)
Classification by
prevention method (PM)
Resource accounting (PM-1)
Classification by
deployment location (DL)
Trained (NBS-2)
Third-party (ADS-3)
Resource multiplication (PM-2)
Source network (DL-3)
Classification by
attack response strategy
(ARS)
Agent identification (ARS-1)
Rate-limiting (ARS-2)
Filtering (ARS-3)
Reconfiguration (ARS-4)
Figure 2: Taxonomy of DDoS Defense Mechanisms
AL-2: Reactive
AL-2:ADS-2:NBS: Normal Behavior Specification
Reactive mechanisms strive to alleviate the impact of an attack on the victim. To attain this goal they need to detect
the attack and respond to it. The goal is to detect every attempted DDoS attack as early as possible and to have a low
degree of false positives. Steps then can be taken to characterize the attack packets and provide this characterization
to the response mechanism.
Based on the specification of a normal behavior, we divide
anomaly detection mechanisms into standard and trained
mechanisms.
AL-2:ADS-2:NBS-1: Standard
Mechanisms that use standard specifications of normal behavior rely on some protocol standard or a set of rules. For
example, the TCP protocol specification describes a threeway handshake that has to be performed for TCP connection setup. An attack detection mechanism can make use of
this specification to detect half-open TCP connections and
delete them from the queue. The advantage of a standardbased specification is that it generates no false positives;
all legitimate traffic must comply to the specified behavior.
The disadvantage is that attackers can still perform sophisticated attacks which, on the surface, seem compliant to the
standard and thus pass undetected.
AL-2:ADS: Attack Detection Strategy
We classify reactive mechanisms based on the attack detection strategy into mechanisms that deploy pattern detection,
anomaly detection, and third-party detection.
AL-2:ADS-1: Pattern Detection
Mechanisms that deploy pattern detection store the signatures of known attacks in a database and monitor each communication for the presence of these patterns. The obvious drawback is that only known attacks can be detected,
whereas new attacks or even slight variations of old attacks
go unnoticed. On the other hand, known attacks are easily
and reliably detected, and no false positives are encountered.
Snort [67] provides one example of a DDoS defense system
that uses pattern attack detection. A similar approach has
been helpful in controlling computer viruses. Like in virus
detection programs, signature databases must be regularly
updated to account for new attacks.
AL-2:ADS-2:NBS-2: Trained
Mechanisms that use trained specifications of normal behavior monitor network traffic and system behavior and generate threshold values for different parameters. All communications exceeding one or more (depending on the approach)
of these values are regarded as anomalous. Trained models catch a broad range of attacks, but have two disadvantages. (1) Threshold setting. Anomalies are detected when
the current system state differs from the model by a certain
threshold. The setting of a low threshold leads to many
false positives, while a high threshold reduces the sensitivity of the detection mechanism. (2) Model update. Systems
and communication patterns evolve with time, and models
need to be updated to reflect this change. Trained specification systems usually perform automatic model update using
statistics gathered at a time when no attack was detected.
This approach makes the detection mechanism vulnerable
to slowly increasing rate attacks that can, over a long period of time, mistrain models and delay or even avoid attack
detection.
AL-2:ADS-2: Anomaly Detection
Mechanisms that deploy anomaly detection have a model
of normal system behavior, such as normal traffic dynamics
or expected system performance. The current state of the
system is periodically compared with the models to detect
anomalies. Approaches presented in [74, 46, 38, 32, 23, 48,
4, 8, 9, 52, 51, 7] provide examples of anomaly detection approaches. The advantage of anomaly detection over pattern
detection is that previously unknown attacks can be discovered. The caveat is that anomaly detectors must trade off
their ability to detect all attacks against their tendency to
misidentify normal behavior as an attack.
ACM SIGCOMM Computer Communications Review
AL-2:ADS-3: Third-Party Detection
Mechanisms that deploy third-party detection do not handle the detection process themselves, but rely on an external
49
Volume 34, Number 2: April 2004
CD-1: Autonomous
message that signals the occurrence of the attack and provides attack characterization. Examples of third-party detection are easily found among traceback mechanisms [10,
62, 25, 66, 65].
Autonomous mechanisms perform independent defense at
the point where they are deployed (a host or a network).
Firewalls and intrusion detection systems provide easy examples of autonomous mechanisms. Even if a defense system
performs its function in a distributed manner it would still
be considered autonomous if it can be completely deployed
within the network it protects (e.g., a network intrusion detection system).
AL-2:ARS: Attack Response Strategy
The goal of the attack response is to relieve the impact of the
attack on the victim while imposing minimal collateral damage to legitimate clients. We classify reactive mechanisms,
based on the response strategy, into mechanisms that deploy
agent identification, rate-limiting, filtering and reconfiguration.
CD-2: Cooperative
Cooperative mechanisms are capable of autonomous detection and response, but can cooperate with other entities and
frequently have significantly better performance in joint operation. The aggregate congestion control (ACC) system
[46] deploying a pushback mechanism [39] provides an example. ACC can autonomously detect the attack, characterize the offending traffic and act locally to impose a rate
limit on that traffic. However, it achieves significantly better performance if the rate-limit requests can be propagated
to upstream routers that otherwise may be unaware of the
attack.
AL-2:ARS-1: Agent Identification
Agent identification mechanisms provide the victim with
(somewhat accurate) information about the identity of the
machines that are performing the attack. This information
can be used by other approaches to alleviate the impact of
the attack. Agent identification examples include numerous
traceback techniques [10, 62, 25, 66, 65]. One frequently
mentioned motivation for deployment of DDoS defenses far
from the victim network is a possible enforcement of liability for attack traffic. A mechanism for reliable and accurate
agent identification would be necessary for liability enforcement.
CD-3: Interdependent
Interdependent mechanisms cannot operate autonomously
at a single deployment point. They either require deployment at multiple networks, or rely on other entities for attack prevention, attack detection or for efficient response.
For example, a traceback mechanism [10, 62, 25, 66, 65]
deployed at a single router would provide no benefit. Secure overlay services [42] are another example of an interdependent mechanism, which needs deployment at multiple
networks and modification of client software.
AL-2:ARS-2: Rate-Limiting
Rate-limiting mechanisms impose a rate limit on a set of
packets that have been characterized as malicious by the
detection mechanism. It is a lenient response technique that
is usually deployed when the detection mechanism has many
false positives or cannot precisely characterize the attack
stream. The disadvantage is that rate limiting will allow
some attack traffic through, so extremely high-scale attacks
might still be effective even if all traffic streams are ratelimited. Examples of rate-limiting mechanisms are found in
[46, 32, 23, 52, 51, 4].
DL: Deployment Location
With regard to deployment location, we differentiate between mechanisms deployed at the victim, intermediate, or
source network.
AL-2:ARS-3: Filtering
Filtering mechanisms use the characterization provided by
detection mechanisms to filter out the attack streams completely. Examples include dynamically deployed firewalls
[24], and some commercial systems [48, 4]. Unless the characterization is very accurate, filtering mechanisms run the
risk of accidentally denying service to legitimate traffic. Worse,
clever attackers might leverage them as denial-of-service tools.
AL-2:ARS-4: Reconfiguration
Reconfiguration mechanisms change the topology of the victim or the intermediate network to either add more resources
to the victim or to isolate the attack machines. Examples include reconfigurable overlay networks [2, 38], resource replication services [74], attack isolation strategies [8], etc.
DDoS defense mechanisms deployed at the victim network
protect this network from DDoS attacks and respond to
detected attacks by alleviating the impact on the victim.
Historically, most defense systems are located at the victim since it suffers the greatest impact of the attack and is
therefore the most motivated to deploy (and bear the cost
of) DDoS defense. Resource accounting [40, 76, 68, 31, 43]
and protocol security mechanisms [44, 50, 5, 63] provide examples of victim network defenses.
DL-2: Intermediate Network
DDoS defense mechanisms deployed at the intermediate network provide infrastructural DDoS defense service to a large
number of Internet hosts. Victims of DDoS attacks can contact the infrastructure and request the service, possibly providing adequate compensation. Pushback [46] and traceback
[10, 62, 25, 66, 65] techniques are examples of intermediatenetwork mechanisms. Such mechanisms are not yet widely
deployed, and many of them can only be effective in wide
deployment.
CD: Cooperation Degree
DDoS defense mechanisms can perform defensive measures
either alone or in cooperation with other entities in the Internet. Based on the cooperation degree, we differentiate
between autonomous, cooperative and interdependent mechanisms.
ACM SIGCOMM Computer Communications Review
DL-1: Victim Network
50
Volume 34, Number 2: April 2004
DL-3: Source Network
nerability (bug or overload). This classification focuses more
on the actual attack phase, while we are interested in looking at the complete attack mechanism in order to highlight
features that are specific to distributed attacks. Hussain
et al. [37] classify flooding (in our terminology brute-force)
DDoS attacks based on number of agent machines performing the attack and whether the attack was reflected or not.
In [35] and [36], Howard proposes a taxonomy of computer
and network attacks. This taxonomy focuses on computer
attacks in general and does not sufficiently highlight features
particular to DDoS attacks. CERT is currently taking the
initiative to devise a comprehensive taxonomy of computer
incidents as part of the design of common incident data format and exchange procedures, but unfortunately results are
not yet available. BBN is also working on generation of a
DDoS attack overview, but its results are not yet released.
The work in [60] provides a valuable discussion of the DDoS
problem and of some defense approaches. A solid body of
work on classification exists in the field of intrusion detection systems [36, 26, 6] and offers informative reading for
researchers in the DDoS defense field.
The goal of DDoS defense mechanisms deployed at the source
network is to prevent network customers from generating
DDoS attacks. Such mechanisms are desirable, but motivation for their deployment is low since it is unclear who would
pay the expenses associated with this service. [32, 23, 52,
51] provide examples of source-network defenses.
6.
USING THE TAXONOMIES
In designing the above taxonomies, we selected those features of attack and defense mechanisms that, in our opinion,
offer critical information regarding seriousness and type of
threats, and effectiveness and cost of defenses. How can
these taxonomies be used?
A map of DDoS research field. For novice researchers,
these taxonomies offer a comprehensive overview for a quick
introduction to the DDoS field. Experienced researchers can
use and extend these taxonomies to structure and organize
their knowledge in the field. This should lead to identification of new directions for DDoS research and improve
understanding of the threat.
Exploring new attack strategies. In addition to known
threats, the attack taxonomy explored a few strategies seen
rarely in the wild (e.g., variable agent set, variable rate attacks), and some novel attack methods (e.g., increasing rate
attacks). As usual suspects get handled by defense systems,
these exotic attacks will gain popularity. Understanding
these threats, implementing them in a test bed environment,
and using them to test defense systems will help researchers
keep one step ahead of the attackers.
DDoS benchmark generation. There is an ongoing effort to
design attack benchmarks for evaluation of DDoS defenses.
The attack taxonomy will help the completeness of benchmark generation. The defense taxonomy will help expose
and identify common weaknesses of a class of DDoS solutions, and design tailored experiments to test these weaknesses.
Common vocabulary. With exception of well understood
attacks (e.g., UDP flood, ICMP flood, etc.), researchers frequently resort to descriptive explanations of sophisticated
attack mechanisms or weaknesses of a specific solution. The
taxonomies offer a common vocabulary for these discussions.
Design of attack class-specific solutions. The aim of DDoS
defenses is frequently to be a silver bullet for all possible attacks. This is unrealistic. The attack taxonomy facilitates
identification of subsets of DDoS threats and design of tailored solutions. If enough such solutions are generated, and
can operate jointly with synergistic effect, we could complete
the puzzle.
Understanding solution constrains. The defense taxonomy highlights common performance constraints and weaknesses for each class of DDoS defenses. Understanding these
problems will focus research efforts on solving them.
Identifying unexplored research areas. Examining the effectiveness of different defense classes against different classes
of attacks should highlight unexplored venues for research.
7.
8.
The DDoS field contains a multitude of attack and defense mechanisms, which obscures a global view of the DDoS
problem. This paper is a first attempt to cut through the
obscurity and structure the knowledge in this field. The proposed taxonomies are intended to help the community think
about the threats we face and the possible countermeasures.
One benefit we foresee from these taxonomies is that of
fostering easier cooperation among researchers. Attackers
cooperate to exchange attack code and information about
vulnerable machines, and to organize their agents into coordinated networks of immense power and survivability. The
Internet community must be equally cooperative within itself to counter the DDoS threat. Good taxonomies will facilitate communication and offer the a common language
for discussing solutions. They will also clarify how different mechanisms are likely to work in concert, and identify
areas of remaining weaknesses that require additional work.
There is a pressing need for the research community to develop common metrics and benchmarks for DDoS defense
evaluation. The taxonomies will be helpful in shaping these
tasks, as well.
The proposed taxonomies are by no means complete and
all-encompassing. New attacks will appear, some of which
we cannot yet imagine. They will highlight new features for
classification. Innovative approaches to DDoS defense will
be designed. They will also offer new design features carrying their share of benefits and weaknesses. We expect these
taxonomies to offer a foundation for classifying threats and
defenses in DDoS field. As the field grows, the taxonomies
will also grow and be refined.
9.
ACKNOWLEDGEMENTS
Authors would like to thank Dr. Sven Dietrich for his
helpful comments, members of the UCLA LASR group for
valuable discussions, and Janice Wheeler for her infinite patience in editing this and many other papers. Verica SavicJovcic generated the customized marking scheme for subsections. This scheme has made the paper infinitely easier to
read, and we are deeply grateful for that.
RELATED WORK
In [41] authors present a classification of denial-of-service
attacks according to the type of the target (e.g., firewall,
Web server, router), a resource that the attack consumes
(network bandwidth, TCP/IP stack) and the exploited vul-
ACM SIGCOMM Computer Communications Review
CONCLUSION
51
Volume 34, Number 2: April 2004
10.
REFERENCES
[1] D. G. Andersen. Mayday: Distributed filtering for
internet services. In In Proceedings of 4th Usenix
Symposium on Internet Technologies and Systems,
March 2003.
[2] D. G. Andersen, H. Balakrishnan, M. F. Kaashoek,
and R. Morris. Resilient Overlay Networks. In
Proceedings of 18th ACM SOSP, October 2001.
[3] T. Anderson, T. Roscoe, and D. Wetherall. Preventing
internet denial-of-service with capabilities. In In
Proceedings of HotNets II, November 2003.
[4] Arbor Networks. The Peakflow Platform.
http://www.arbornetworks.com.
[5] T. Aura, P. Nikander, and J. Leiwo. DOS-Resistant
Authentication with Client Puzzles. Lecture Notes in
Computer Science, 2133, 2001.
[6] S. Axelsson. Intrusion detection systems: A survey
and taxonomy. Technical Report 99-15, Department of
Computer Engineering, Chalmers University, March
2000.
[7] P. Barford, J. Kline, D. Plonka, and A. Ron. A signal
analysis of network traffic anomalies. In In Proceedings
of the 2nd ACM SIGCOMM Internet Measurement
Workshop, November 2002.
[8] BBN Technologies. Applications that participate in
their own defense.
http://www.bbn.com/infosec/apod.html.
[9] BBN Technologies. Intrusion tolerance by
unpredictability and adaptation.
http://www.bbn.com/infosec/itua.html.
[10] S. Bellovin, M. Leech, and T. Taylor. ICMP
Traceback Messages. Internet draft, work in progress,
October 2001.
[11] D. J. Bernstein. Syn cookies.
http://cr.yp.to/syncookies.html.
[12] CERT CC. CA-2001-19 “Code Red” Worm Exploiting
Buffer Overflow In IIS Indexing Service DLL.
http://www.cert.org/advisories/CA-2001-19.html.
[13] CERT CC. Code Red II.
http://www.cert.org/incident notes/IN-2001-09.html.
[14] CERT CC. Denial of Service Attacks.
http://www.cert.org/tech tips/denial of service.html.
[15] CERT CC. DoS using nameservers.
http://www.cert.org/incident notes/IN-2000-04.html.
[16] CERT CC. erkms and li0n worms.
http://www.cert.org/incident notes/IN-2001-03.html.
[17] CERT CC. Nimda worm.
http://www.cert.org/advisories/CA-2001-26.html.
[18] CERT CC. Ramen worm.
http://www.cert.org/incident notes/IN-2001-01.html.
[19] CERT CC. Smurf attack.
http://www.cert.org/advisories/CA-1998-01.html.
[20] CERT CC. TCP SYN flooding and IP spoofing attacks.
http://www.cert.org/advisories/CA-1996-21.html.
[21] CERT CC. Trends in Denial of Service Attack
Technology, October 2001.
http://www.cert.org/archive/pdf/DoS trends.pdf.
[22] Cisco. Strategies to protect against Distributed Denial
of Service Attacks.
http://www.cisco.com/warp/public/707/newsflash.html.
[23] Cs3. Inc. MANAnet DDoS White Papers.
ACM SIGCOMM Computer Communications Review
52
http://www.cs3-inc.com/mananet.html.
[24] T. Darmohray and R. Oliver. Hot spares for DDoS
attacks.
http://www.usenix.org/publications/login/20007/apropos.html.
[25] D. Dean, M. Franklin, and A. Stubblefield. An
algebraic approach to IP Traceback. In Proceedings of
the 2001 Network and Distributed System Security
Symposium, February 2001.
[26] H. Debar, M. Dacier, and A. Wespi. Towards a
taxonomy of intrusion-detection systems. In Computer
Networks, volume 31(8), pages 805–822, April 1999.
[27] D. Dittrich. The DoS Project’s trinoo distributed
denial of service attack tool.
http://staff.washington.edu/dittrich/misc/
trinoo.analysis.
[28] D. Dittrich. The Tribe Flood Network distributed
denial of service attack tool.
http://staff.washington.edu/dittrich/misc/
tfn.analysis.txt.
[29] D. Dittrich, G.Weaver, S. Dietrich, and N.Long. The
mstream distributed denial of service attack tool.
http://staff.washington.edu/dittrich/misc/
mstream.analysis.txt.
[30] P. Ferguson and D. Senie. Network Ingress Filtering:
Defeating Denial of Service Attacks which Employ IP
Source Address Spoofing. RFC 2827, May 2000.
[31] A. Garg and A. L. N. Reddy. Mitigation of DoS
attacks through QoS Regulation. In Proceedings of
IWQOS workshop, May 2002.
[32] T. M. Gil and M. Poletto. MULTOPS: a
data-structure for bandwidth attack detection. In
Proceedings of 10th Usenix Security Symposium,
August 2001.
[33] K. Hafner and J. Markoff. Cyberpunk: Outlaws and
hackers on the computer frontier. Simon & Schuster,
1991.
[34] G. Hardin. The Tragedy of the Commons. Science,
162(1968):1243–1248, 1968.
[35] J. D. Howard. An analysis of security incidents on the
Internet. PhD thesis, Carnegie Mellon University,
August 1998.
[36] J. D. Howard and T. A. Longstaff. A common
language for computer security incidents.
[37] A. Hussain, J. Heidemann, and C. Papadopoulos. A
Framework for Classifying Denial of Service Attack. In
Proceedings of SIGCOMM 2003, 2003.
[38] Information Sciences Institute. Dynabone.
http://www.isi.edu/dynabone/.
[39] J. Ioannidis and S. M. Bellovin. Pushback:
Router-Based Defense Against DDoS Attacks. In
Proceedings of NDSS, February 2002.
[40] A. Juels and J. Brainard. Client puzzles: A
cryptographic countermeasure against connection
depletion attacks. In Proceedings of the 1999 Networks
and distributed system security symposium, March
1999.
[41] F. Kargl, J. Maier, and M. Weber. Protecting web
servers from distributed denial of service attacks. In
Proceedings of 10th International World Wide Web
Conference, May 2001.
Volume 34, Number 2: April 2004
[42] A. D. Keromytis, V. Misra, and D. Rubenstein. SOS:
Secure Overlay Services. In Proceedings of SIGCOMM
2002, 2002.
[43] F. Lau, S. H. Rubin, M. H. Smith, and L. Trajkovic.
Distributed Denial of Service Attacks. In IEEE
International Conference on Systems, Man, and
Cybernetics, pages 2275–2280, Nashville, TN, USA,
October 2000.
[44] J. Leiwo, P. Nikander, and T. Aura. Towards network
denial of service resistant protocols. In Proceedings of
the 15th International Information Security
Conference, August 2000.
[45] J. Li, J. Mirkovic, M. Wang, P. Reiher, and L. Zhang.
SAVE: Source Address Validity Enforcement Protocol.
In Proceedings of INFOCOM 2002, June 2002. to
appear.
[46] R. Mahajan, S. Bellovin, S. Floyd, V. Paxson, and
S. Shenker. Controlling high bandwidth aggregates in
the network. ACM Computer Communications
Review, 32(3), July 2002.
[47] G. R. Malan, D. Watson, F. Jahanian, and P. Howell.
Transport and Application Protocol Scrubbing. In
Proceedings of INFOCOM 2000, pages 1381–1390,
2000.
[48] Mazu Networks. Mazu Technical White Papers.
http://www.mazunetworks.com/white papers/.
[49] McAfee. Personal Firewall.
http://www.mcafee.com/myapps/firewall/ov firewall.asp.
[50] C. Meadows. A formal framework and evaluation
method for network denial of service. In Proceedings of
the 12th IEEE Computer Security Foundations
Workshop, June 1999.
[51] J. Mirkovic. D-WARD: Source-End Defense Against
Distributed Denial-of-Service Attacks. PhD thesis,
University of California Los Angeles, August 2003.
[52] J. Mirkovic, G. Prier, and P. Reiher. Attacking DDoS
at the Source. In Proceedings of the ICNP 2002,
November 2002.
[53] D. Moore. The spread of the code red worm (crv2).
http://www.caida.org/analysis/security/codered/coderedv2 analysis.xml.
[54] D. Moore, G. Voelker, and S. Savage. Inferring
Internet Denial-of-Service Activity. In Proceedings of
the 2001 USENIX Security Symposium, 2001.
[55] R. Naraine. Massive DDoS Attack Hit DNS Root
Servers, October 2002.
http://www.esecurityplanet.com/trends/
article/0,,10751 1486981,00.html.
[56] National Infrastructure Protection Center. Advisory
01-014: New Scanning Activity (with
W32-Leave.worm) Exploiting SubSeven Victims, June
2001.
http://www.nipc.gov/warnings/advisories/2001/01014.htm.
[57] E. O’Brien. NetBouncer : A practical client
legitimacy-based DDoS defense via ingress filtering.
http://www.nai.com/research/nailabs/developmentsolutions/netbouncer.asp.
[58] K. Park and H. Lee. On the Effectiveness of
Route-Based Packet Filtering for Distributed DoS
Attack Prevention in Power-Law Internets. In
ACM SIGCOMM Computer Communications Review
53
Proceedings of ACM SIGCOMM 2001, August 2001.
[59] V. Paxson. An analysis of using reflectors for
distributed denial-of-service attacks. ACM Computer
Communications Review (CCR), 31(3), July 2001.
[60] V. Razmov. Denial of Service Attacks and How to
Defend Against Them.
http://www.cs.washington.edu/homes/valentin/
papers/DoSAttacks.pdf.
[61] SANS Institute. NAPTHA: A new type of Denial of
Service Attack, December 2000.
http://rr.sans.org/threats/naptha2.php.
[62] S. Savage, D. Wetherall, A. Karlin, and T. Anderson.
Practical Network Support for IP Traceback. In
Proceedings of ACM SIGCOMM 2000, August 2000.
[63] C. Schuba, I. Krsul, M. Kuhn, G. Spafford,
A. Sundaram, and D. Zamboni. Analysis of a denial of
service attack on TCP. In Proceedings of the 1997
IEEE Symposium on Security and Privacy, May 1997.
[64] S.Dietrich, N. Long, and D. Dittrich. An Analysis of
the ”shaft” distributed denial of service tool. In
Proceedings of LISA 2000, 2000.
http://www.adelphi.edu/ spock/shaft-lisa2000.pdf.
[65] A. C. Snoeren, C. Partridge, L. A. Sanchez, C. E.
Jones, F. Tchakountio, S. T. Kent, and W. T. Strayer.
Hash-Based IP Traceback. In Proceedings of ACM
SIGCOMM 2001, August 2001.
[66] D. X. Song and A. Perrig. Advanced and
authenticated marking schemes for IP Traceback. In
Proceedings of IEEE Infocom 2001, 2001.
[67] Sourcefire. Snort: The Open Source Network Intrusion
Detection System.
[68] O. Spatscheck and L. L. Petersen. Defending Against
Denial of Service Attacks in Scout. In Proceedings of
the 3rd Symposium on Operating Systems Design and
Implementation, February 1999.
[69] S. Staniford, J. Hoagland, and J. McAlerney. Practical
automated detection of stealthy portscans. Journal of
Computer Security, 10(1/2), 2002.
[70] S. Staniford, V. Paxson, and N. Weaver. How to 0wn
the internet in your spare time, 2002. In Proceedings
of the 11th USENIX Security Symposium.
[71] Tripwire. Tripwire for servers.
http://www.tripwire.com/products/servers/.
[72] N. Weaver. Warhol Worm.
http://www.cs.berkeley.edu/ nweaver/worms.pdf.
[73] M. Williamson. Throttling viruses: Restricting
propagation to defeat malicious mobile code. In 18th
Annual Computer Security Applications Conference,
December 2002.
[74] J. Yan, S. Early, and R. Anderson. The XenoService –
A Distributed Defeat for Distributed Denial of
Service. In Proceedings of ISW 2000, Oct. 2000.
[75] V. Yegneswaran, P. Barford, and J. Ullrich. Internet
intrusions: Global characteristics and prevalence. In
In Proceedings of the 2003 ACM SIGMETRICS
International conference on Measurement and
Modeling of Computer Systems, pages 138–147, 2003.
[76] Y. L. Zheng and J. Leiwo. A Method to Implement a
Denial of Service Protection Base. In Information
Security and Privacy, volume 1270 of LNCS, pages
90–101, 1997.
Volume 34, Number 2: April 2004
ACM SIGCOMM Computer Communications Review
54
Volume 34, Number 2: April 2004