See discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/221632294
A Terminology for Control Models at Optical
Exchanges
Conference Paper · June 2007
DOI: 10.1007/978-3-540-72986-0_5 · Source: DBLP
CITATIONS
READS
7
16
8 authors, including:
Freek Dijkstra
Leon Gommans
38 PUBLICATIONS 343 CITATIONS
53 PUBLICATIONS 587 CITATIONS
SURFsara
SEE PROFILE
Air France KLM
SEE PROFILE
Jeroen Van der Ham
University of Amsterdam
48 PUBLICATIONS 327 CITATIONS
SEE PROFILE
All content following this page was uploaded by Freek Dijkstra on 12 December 2016.
The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document
and are linked to publications on ResearchGate, letting you access and read them immediately.
A Terminology for Control Models at Optical
Exchanges
Freek Dijkstra1 , Bas van Oudenaarde2 , Bert Andree1 , Leon Gommans1 , Paola
Grosso1 , Jeroen van der Ham1,3 , Karst Koymans1 , Cees de Laat1
1
Universiteit van Amsterdam, Kruislaan 403, Amsterdam, The Netherlands
fdijkstr@science.uva.nl
2
Finalist IT Group, Postbus 1354, Rotterdam, The Netherlands
3
TNO Defense, Security and Safety, The Hague, the Netherlands
Abstract. Optical or lambda exchanges have emerged to interconnect
networks, providing dynamic switching capabilities on OSI layer 1 and
layer 2. So far, the only inter-domain dynamics have occurred on layer 3,
the IP layer. This new functionality in the data plane has consequences
on the control plane. We explain this by comparing optical exchanges
with current Internet exchanges.
Descriptions of optical exchanges have appeared in the literature, but
discussions about these exchanges have been hampered by a lack of
common terminology. This paper defines a common terminology for exchanges. Discussion in the community revealed four different meaning
for the term “open exchange”. We list them in this paper.
We classify the different kind of exchanges based on the interactions
between the domains at the control plane. We use these control models
to distinguish between different types of interconnection points.
1
1.1
Introduction
Overview
The main function of Interconnection points, such as exchanges, is to facilitate
traffic flows between the connected domains. Besides regular Internet-based exchanges, new types of exchanges are emerging. A wide variety of names has
been proposed for these new exchanges, including optical exchange, transport
exchange, grid exchange, GLIF open lambda exchange (GOLE), optical interconnection point and lightpath exchange.
The goal of this paper is to create a generally usable terminology for exchanges, both optical and Internet exchanges. The novelty in our work comes
from the fact that we do so by looking at the control plane rather than the
data plane, we identified conflicting definitions, and we are the first to compare
optical and internet exchanges in detail.
Section 2 gives a classification of existing and new exchanges, and defines
our terminology. Where possible, existing terminology is re-used. Other termi-
nology, in particular the term open exchange, draws upon discussions in the
GLIF community [1,2]4 .
A distinguishing factor for exchanges is the ability or inability of connected
domains to influence the state of the core network. To this end, we define a total
of three control models for exchanges in Sect. 3. This categorization will aid
the discussion about the design of new exchanges. Section 4 maps these control
models to each type of exchange.
The paper concludes with future trends and conclusions.
We refer to the extended version of this paper for a discussion about advanced
network services on the data plane, like the conversion of data between different
formats (interworking) and layers (elevator services), or on the control plane, like
automated provisioning of network elements, policy based authorization, broker
services, and index servers [3,4].
1.2
Related Work
This work builds on experience and discussions in the GLIF community, a collaboration of mostly National Research and Education Networks (NRENs). Here
the need for high bandwidth circuits led to hybrid networks offering both routed
and circuit switched network connections. Interconnections between NRENs are
often made at new optical exchanges, like NetherLight, StarLight, ManLan, TLex, HK Light, UKLight and NorthernLight.
We rely as much as possible on exisiting terminology. In particular, the ownership terminology in Sect. 2.3 builds upon the management layers in Telecommunication Management Network (TMN) [5] and current practice in economic
and legal communities [6].
This paper deals with the network node interface (NNI) of networks connected to an exchange, and is by no means the first to discuss this interface. The
Optical Interworking Forum specified the network to network interface between
domains (E-NNI) based on RSVP messaging [7]. Recent work comes from the
L1VPN [8] workgroup in the IETF, which deals with the NNI for GMPLS [9].
The work provided in this paper is complimentary because it specifically
deals with the network interface for an exchange rather than a transit network.
This paper deals with a high level overview of the relation between the different
actors, rather than specifying a practical signaling protocol.
2
Terminology
In this section we introduce a concise definition of terms like domain, administrative control, as well as open and automated.
4
The only exception is that we use the term “optical exchange”. The GLIF community
currently uses the term “GOLE”, and the authors personally prefer the term “transport exchange”, but we felt that “optical exchange” was most widely recognized in
all communities.
2.1
Peering
Traffic between separate networks is often exchanged at geographically clustered
locations, called interconnection points or peering points [10,11]. For the regular Internet, the Internet service providers (ISPs), can interconnect using either
transit or peering [12]. Peering, in most literature, is limited to providing connectivity to each others networks and to the customers of the other network, but
not to other destinations. Transit on the other hand implies that traffic for any
destination can be handled by the party providing the connectivity, usually for
a fee.
In this paper we do not distinguish between peering and transit. In our
terminology peers are network owners who connect to an interconnection point
and peering is the concept of exchanging traffic between peers, regardless of
the economic model.
2.2
Types of Interconnection Points
The most trivial interconnection point is a co-location that only provides rack
space and power. This already gives the ability to initiate bilateral peerings
between peers at the same facility. We are interested in exchanges, which are
interconnection points with one or more core networks in place, dedicated to the
exchange of traffic between peers.
Classification. We currently observe four types of interconnection points, based
on the function, rather than the technical implementation:
–
–
–
–
Internet exchanges
mobile roaming exchanges
optical exchanges
points of presence
Internet exchanges, also known as Internet exchange points (IXP) or Network
access points (NAP), serve as an interconnection points to exchange packet data
between individual peers. The peers have one or a few physical connections to
a central core infrastructure. The core network can be Ethernet LAN, ATM, or
MPLS-based. The first variant is stateless, while the other two are stateful and
require that the individual peers set up a path between them. Such a path is a
channel in the physical connection.
Mobile roaming exchanges, such as GPRS roaming exchanges (GRX)
[13] and UMTS exchanges, exchange packet data for respectively 2.5th and 3rd
(3G) generation mobile telephony. In telecommunications, however, the term
“exchange” is different from our usage and refers to a transit provider rather than
an interconnection point. An exchange point between mobile roaming exchanges
is technically not different from a packet-based5 Internet exchange.
5
GPRS and UMTS are packet based. The older CSD system is circuit switched.
Optical exchanges6 , also known as lambda exchanges, grid exchange points,
transport exchanges or GLIF open lambda exchanges, are interconnection points
where peers exchange traffic at OSI layer 1 or layer 2 [3]. GMPLS Internet exchanges as defined by Tomic and Jukan [14] share the concept of circuit-switched
interconnection points, but have not been implemented yet.
We use the term Transport Exchange to refer to circuit-switched exchanges, like current-day optical exchanges.
Unlike exchanges, points of presence (POP) are interconnection points
where the peers are unequal. Access networks connect with an upstream network
provider at a POP. In this case, the peers are unequal since the upstream provider
accepts transit traffic from the customer, but the reverse is not true.
Internet versus Optical Exchanges. Table 1 highlights the differences between Internet exchanges and optical exchanges. Peers at an Internet exchange
interconnect to exchange IP traffic. The core of an Internet exchange contains
exactly one circuit per peering relation. In contrast, an optical network supports
circuits between end-users, so at an optical exchange there is a circuit between
peers for each end-to-end connection that goes through the exchange. The table
further emphasizes that for an optical exchange these circuits can carry any layer
1 or layer 2 traffic. Differences in function and purpose lead to different choices
in technology between Internet exchanges and optical exchanges. Finally, the
table highlights that an optical exchange may offer more advanced services than
an Internet exchange.
Table 1. Functional differences between Internet exchanges and current optical
exchanges.
Internet Exchange
OSI Layer Transports traffic at layer 2, peers
connect with layer 3 devices
Traffic
IP traffic only
type
End-points Connection between two peering
networks
Dynamics Stateless, or state changes only
when peering relations change
Technology Often packet switched, sometimes
label-switched (with virtual circuits like MPLS and ATM)
Services
Only data transport
6
Optical Exchange
Transports traffic at layer 1 or layer
2, peers connect at that same layer.
Any packet data or any data at a
specific framing or bit rate
Connections are part of a larger circuit between two end-hosts
State changes for each data transport
Circuit or virtual-circuit switched
(e.g. using SONET or VLANs)
Data transport and other services,
like the conversion of data between
different formats and layers
Optical does not imply that the exchange itself is purely photonic.
There is no clear boundary between the different interconnection points since
each interconnection point may take multiple roles. We expect that the differences listed in Table 1 will change over time, as new technologies become available and are implemented. For example, customers at a POP may also directly
peer with each other, a function typically seen at exchanges. Circuit switching is
typically associated with optical exchanges, but not a technical necessity: ATMand MPLS-based Internet exchanges are also circuit switched and it might be
possible to create a non-circuit switched optical exchange using optical burst
switching (OBS) [15].
2.3
Ownership
Owner, Administrator and Users. We distinguish between legal owner, economic owner, administrator and user(s) for each network element7 . The legal
owner of a network element is the entity that purchased the device and the economic owner is the entity that acquired the usage rights from the legal owner.
We base these terms on current practice in economic and legal communities [6].
The economic owner determines its policy of the network. This entity carries the responsibility for the behavior of a device and has the final responsibility
in case of hazards and abuse. In addition, each network element can also have a
separate administrator, the person, organization, or software component that
configures and administers the device on behalf of the economic owner. The
economic owner determines the policy for a network element; the administrator
enforces this policy. Finally, the users may use or invoke an element, if their
request is in compliance with the active policy.
We assume that each network element has exactly one legal owner, one economic owner, and one administrator, but may have multiple users over time
(though typically only one at a specific time).
Domains. We define a domain as a set of network elements8 . An administrative domain is a set of network elements with the same administrator. An
owner domain is a set of network elements with the same economic owner.
A core network is an administrative domain within an interconnection point
that is able to exchange traffic between at least three peers. Core networks are
of special interest throughout this paper and we use the term core to refer to a
core network and its administrator.
Examples. Often the legal owner, economic owner, and administrator of a
network element are the same entity. For example, in the Internet, a transit
provider is typically owner and administrator of its network. But this is not
always the case.
7
8
Network element is a generic term to include network devices, links, interfaces and
hosts.
Including non-disjoint sets. Note that a domain does not necessarily have to be an
AS-domain.
An organization leases a trans-oceanic fiber from a carrier for a year, the
carrier is the legal owner, while the other organization is the economic owner.
If an organization outsources the maintenance of its network, the economic
owner and administrator of this network are different entities.
In the next subsection we explain the concept of open control, where the exchange is both the legal owner as well as the administrator of a specific interface,
while the peer is the economic owner of this interface.
2.4
Open Exchanges
We found that in the the GLIF community, the use of “open” in “open exchanges” was ambiguous. It could refer to at least four different meanings, as
described below. We recommend that it is only used in the now prevalent meaning of open control. For other meanings, we suggest alternative wording.
Open Control Model. In a closed interconnection point, the economic owner
domain is equal to the administrative domain: everyone both decides upon and
enforces the policy of their network elements. In particular, the core ultimately
decides on the policy for each interface in the core network.
In the open control model, the core of an open exchange delegates the policy
decision of each external interface to the peer that connects to that interface.
Therefore, peers of an open exchange have the ability to configure “their” interfaces in the core network and thus can decide who connects to their networks.
Figure 1 shows an optical exchange consisting of an optical cross connect at
the core. The exchange has three peers: Anet, Bnet and Cnet. If Anet wants
to connect to Cnet, it signals that request to the exchange. A closed exchange
would autonomously decide to grant or deny that request, and choose interface 4
or 5. An open exchange outsources this policy decision to Cnet which has policy
control over interface 4 and 5, even though this policy is enforced in the optical
cross connect which is legally owned and administrated by the exchange.
Anet
Anet
1
2
1
4
2
4
3
5
Cnet
3
Bnet
5
Cnet
Bnet
Fig. 1. Example of an optical exchange. On the left the administrative domains
are shown, which are equal to the owner domains for the closed control model.
On the right, the owner domains for the open control model are shown.
In the open control model, the core does not define an acceptable use policy
(AUP) for its peers, and is thus AUP free.
Business Model. We use the word “public” or “neutral” to refer to an
interconnection point with an open business model. An open business model
requires that an interconnection point must have a published, clear policy for
new peers to join, and has a reasonable and non-discriminatory (RAND) policy9
towards its peers.
A non-public interconnection point is called “private” or “non-neutral”.
An open exchange can still be non-neutral. For example, an exchange economic
owner may decide to only connect business partners as peers, but not others,
and have the partners then decide on the policy for connections. Similarly, a
neutral exchange may not be open. Hypothetically, an exchange may decide to
allow every peer to connect to the core network, but grant path setup requests
depending on an arbitrary decision.
Service Exposure. The term “service exposure” can be used to refer to
the ability by peers to look in the inner workings of the exchange. The opposite
of service exposure is “service overlay”. An exchange with a service overlay
would behave like a black box. While peers can make requests to a black box,
they do not know what exact devices, interfaces or other equipment are used to
fulfill the request.
Automated Exchange. An exchange is called “automated” if peers are able
to set up circuits between each other and invoke other services from the exchange
without manual intervention from the economic owner of the core network.
3
Control Models
In this section, we define three different control models for interconnection points:
the autonomous, federated and distributed control models. The autonomous
control model is the simplest model. The federated and the distributed control
model respectively extend the autonomous and the federated control models.
These models make a clear distinction between administrative control (policy
enforcement) and owner control (policy decision) of the network elements. We
consider a few administrative domains on the transport plane, each operated by
a specific administrator. For each model, we explain how owner domains control
network elements, and in particular how peers decide on the business policy for
some network elements in the core network.
It is only possible to control network elements in another administrative
domain if the administrators work together by sending messages to each other.
It should be noted that we do not assume that these messages are automated.
9
This may seem to imply equal access rights to all peers. However, a distinction can
be made based on the service level, as long as the service level is achievable by all
peers on non-discriminatory conditions. E.g., if they pay a certain fee.
3.1
Autonomous Control Model
In the autonomous control model, there is exactly one core network, which is
owned and administrated by a single entity. Peers can connect their network to
the interconnection point, but there is no interaction between the peers and the
core network on the control plane. Peers may interact with each other, but that
is not relevant to this model.
Figure 2 shows an example of the autonomous control model. In this figure,
the transport plane shows five distinct administrative domains: core, A, B, C
and D, each operated by a administrator on the control plane. On the transport
plane, each box represents an administrative domain, interconnected by links.
On the control plane, each square represents a separate controller. There is no
communication between the peers and the core on the control plane.
Control plane
Transport
plane
B
A
core
C
D
Fig. 2. Example of the autonomous control model. Squares represent administrative domains.
The economic owner of a core network determines a limited number of policies. Peers either accept the policies or take their business elsewhere.
The peers of a LAN-based Internet exchanges exchange control messages
using an external routing protocol, but not with the exchange itself. So these
exchanges are examples of the autonomous control model.
The autonomous control model is always closed.
3.2
Federated Control Model
In the federated control model, the interconnection point has exactly one core
network. The core offers services to each peer, including the ability to interconnect with other peers.
The inner workings of the core network may be unknown to the peers (making
it a black box), but peers can still check information about the state of some
resources. For example, a peer can still inquire about the availability of a certain
resource or get the status of a circuit it established earlier.
Figure 3 shows an example of the federated control model. The transport
plane is the same as in Fig. 2, but the control plane is different: here the controller
of each peer exchanges messages with the controller of the core network.
Control plane
Transport
plane
B
A
core
C
D
Fig. 3. Example of the federated control model.
When a peer wants to use a certain service, it invokes the administrator of
the core network, which may delegate parts of the request to other parties. For
example, if peer D sends a request to set up a circuit from B to D, the core
economic owner checks if the requested resources in the core itself are available
and contacts the economic owner(s) of the resources involved. In the case of open
control, the core asks peer B if this request must be honored. If that is true, the
core administrator then creates the requested circuit.
3.3
Distributed Control Model
In the distributed control model there can be multiple federations, each controlling a different core network. Every party can bring in its own equipment, e.g.
fibers, and most important: its own services (and control software). Each peer
exposes its own services to the rest of the community, possibly without revealing the inner details. A broker may combine multiple services and expose this
combination as a single service.
The idea is that each peer still administratively controls its own network
elements, but interacts with other administrators, or partially delegates its policy
control, forming collaborations. Each peer can partner in multiple collaborations.
It is possible to regard one instance of the distributed control model as multiple interconnected instances of the federated control model. However, the distributed control model highlights the intelligence that is required to make all
parts work together. This intelligence is not always necessary in the federated
model.
Figure 4 shows an example of the distributed control model. The figure shows
how peers can dedicate part of their network resources to form a dedicated core
network. For example, A may expose some network elements to the other peers,
which can be used by B or D to interconnect, either to A, or between each
other through the core network of A. Also, C and D may decide to put some
network resources in a pool, forming another, joint, core network. Typically, a
core network formed by multiple peers is exposed as one single core network by a
broker, which then delegates incoming requests to the individual administrators
of the peers.
Control plane
Y
C
Transport
plane
B
A
A
D
B
core X
core Y
C
D
Fig. 4. Example of the distributed control model.
4
Model Usage
In Table 2, we give a list of viable mappings between the current interconnection
points to the models described, based on our observation of current exchanges.
Table 2. Applicable models for each type of interconnection point.
Autonomous control model
Federated control model
Distributed control model
Internet
Exchange
X
X
Mobile
Exchange
X
X
Optical
Exchange
X
X
Point of
Presence
X
X
Stateless Internet and mobile exchanges use the autonomous control model,
since no request needs to be sent to the core network administrator to exchange
data between peers. If the Internet or mobile exchange is stateful, it can be either
of these two models.
A POP typically uses the autonomous control model, because the configuration is mostly static and peers have no direct control over the inner working of
the facility. However, if peers of a POP can decide on the policy, the federated
control model is used.
If optical exchanges offer multiple services, standardized service discovery
and service invocation are required. Both the federated and distributed control
models offer this feature in a scalable way using pluggable services (a service
oriented architecture). The distributed control model is more complex than the
other models, and thus harder to deploy, because there is no longer a single
entity that acts as a broker broker.
5
Future Trends
Large data transport on long distances is most efficient over the lowest possible
layers, and peers and their users demand more flexibility to set up circuits with
known quality of service (QoS) between domains. Interconnection points down
in the protocol stack can offer this flexibility.
Technologies change over time, just as the requests from the users. We have
reasons to believe that the current optical (transport) exchanges and Internet
exchanges converge into optical exchanges that support all the required services.
First there is a tendency for current optical exchanges to provide network services, and a future service might be multiparty peering like in a LAN-based Internet exchange. Secondly, Internet exchanges tend to offer more services which
are now regarded as optical exchange functions, like private circuits between two
peers10 . Third, there is a tendency to build Internet exchanges and optical exchanges at the same locations11 , which indicates a possible economic advantage
of combining exchanges on the same physical location.
Open control is a mind shift compared to most current exchanges. With
closed control, peers sometimes have the ability to change the state of one or more
network elements in a core network, but their requests are evaluated against the
policy set by the exchange. With open control on the other hand, the peers decide
on the policy and the exchange enforces it for them. Even if peers are in control,
they do not experience it that way unless their requests are promptly answered
by an automated ensemble. Thus, automation of exchanges is a necessity for this
paradigm change to happen.
We also recognize a trend to let end users control the network resources as
they want. For example UCLP supported by CANARIE is a control mechanism
driven by users. Whether the exposition of network elements and network services will continue is yet unclear. If low layer network connections are exposed
to users, authorization becomes more important to prevent abuse. Monitoring is
important for peers and end-users to check if and where failures occur. This is
part of our future research direction.
6
Conclusion
Formerly, discussions about optical or lambda exchanges have been hampered by
a lack of common terminology. In this paper we identified ambiguous terms, in
particular on “open exchanges”, and presented a consistent terminology, based
on experiences in the GLIF community. We introduced multiple models for exchanges that we offer to use as reference points to the community. We did show
that the terminology can be used to classify the existing exchanges according
to the models that we introduced. While we are confident that the models are
workable, we hope they are found as fruitful to others as they are to use in
discussions on the difference between Internet exchanges and optical exchanges.
Acknowledgment
Part of this research is done under the GigaPort Next Generation project led by
the Dutch National Research Network (SURFnet), and the Interactive Collabo10
11
For example, the Amsterdam Internet Exchange AMS-IX already provides private
interconnects and closed user groups.[16]
For example, Chicago, New York and Amsterdam.
rative Information Systems (ICIS) project. Both projects are supported by the
Dutch Ministry of Economic Affairs, grant numbers BSIK03020 and BSIK03024.
The authors wish to thank John Vollbrecht from Internet2, Henk Steenman
and Job Witteman of the AMS-IX, and members of the GLIF community for
their discussions and proof-reading.
References
1. Global Lambda Integrated Facility, http://www.glif.is/
2. Terminology discussion in the GLIF community, Sep. 2005, http://www.glif.is/
list-archives/tech/msg00019.html
3. F. Dijkstr, C.T.A.M. de Laat, “Optical Exchanges”, GRIDNETS conference proc.,
Oct. 2004
4. F. Dijkstra, et al, extended version of “A Terminology for Control Models at Optical
Exchanges”, http://www.science.uva.nl/∼fdijkstr/
5. “Considerations for a telecommunications management network”, ITU recommendation M.3013, Feb. 2000
6. The terms “economic ownership” and “legal ownership” are common economic
terms. E.g. Anne Harrison in “Definition of economic assets”, Jan. 2006, http:
//unstats.un.org/UNSD/nationalaccount/AEG/papers/m4EconAssets.pdf
7. Lyndon Y. Ong (ed), et al., “Intra-Carrier E-NNI Signaling Specification”, OIF
specification OIF-E-NNI-Sig-01.0, Feb. 2004
8. Tomonori Takeda, et al., “Framework and Requirements for Layer 1 Virtual Private
Networks”, draft-ietf-l1vpn-framework, Jan. 2007, Work in Progress
9. Lou Berger, et al., “Generalized Multi-Protocol Label Switching (GMPLS) Signaling
Functional Description”, RFC 3471, Jan. 2003
10. Bilal Chinoy and Timothy Salo, “Internet Exchanges: Policy-Driven Evolution”,
Harvard Workshop On Co-Ordination Of The Internet, J.F. Kennedy School Of Government, Sep. 1996
11. Geoff Huston, “Interconnection, Peering, and Settlements”, Proc. of Inet’99,
Jun. 1999
12. William Norton, “Internet Service Providers and Peering”, Proc. of NANOG 19,
May 2001
13. K.J. Blyth, A.R.J. Cook, “Designing a GPRS roaming exchange service”, Second
International Conference on 3G Mobile Communications Technologies, Mar. 2001
14. Slobodanka Tomic, Admela Jukan, “GMPLS-Based Exchange Points: Architecture
and Functionalilty”, Chapter 8 in “Emerging Optical Network Technologies Architectures, Protocols and Performance”, Springer, ISBN 0-387-22582-X, Oct. 2004
15. Chunming Qiao, Myungsik Yoo, “Optical Burst Switching (OBS) – A New
Paradigm for an Optical Internet”, J. of High-Speed networks, pp. 69-84, 1999
16. Amsterdam Internet Exchange, “Services provided by the AMS-IX”, http://www.
ams-ix.net/services/