Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                

Bootcamp MySQL HeatWave

Download as pdf or txt
Download as pdf or txt
You are on page 1of 45

Tech: Project Kit

Applications with MySQL HeatWave


Technical Presentation

Alexandre Fagundes
alexandre.af.fagundes@oracle.com
Cloud Architect, Oracle Latin America
MySQL: An extremely popular database

Copyright © 2024, Oracle and/or its affiliates


MySQL is the #1 Open Source Database

Copyright © 2024, Oracle and/or its affiliates


MySQL is the most popular database for developers

Which databases have you used in the last 12 months?

MySQL 51%

PostgreSQL 38%

MongoDB 27%

Redis 26%

SQLite 25%

MS SQL Server 18%

Jetbrains survey 2023

Copyright © 2024, Oracle and/or its affiliates


Innovative organizations across many industries run MySQL – Do you?

Social E-Commerce Tech Finance Manufacturing

Copyright © 2024, Oracle and/or its affiliates


MySQL powers Open Source applications

Custom Apps Development Content management and Learning platforms


eCommerce

Copyright © 2024, Oracle and/or its affiliates


Challenges for developers and DBAs…
…And how MySQL HeatWave uniquely addresses them

Copyright © 2024, Oracle and/or its affiliates


What we’ve heard from customers

We provide one MySQL cloud database service for


Complex and costly to use separate
OLTP and real-time analytics across data warehouses
systems for transactions and analytics
and data lakes—without ETL duplication

We provide automated in-database ML with an


Want to leverage ML and generative AI explanation of models and results, without ETL, plus
on all their data generative AI with Vector store (private preview)

We provide a fully managed database service with


Spend too much time on manual
machine learning-powered automation and built-in
management tasks
advanced security features

Want to use multiple clouds MySQL HeatWave is available on OCI, AWS, Azure, and
in your data center

Copyright © 2024, Oracle and/or its affiliates


MySQL is optimized for OLTP, not designed for analytic processing

OLTP DB

Separate analytics database


OLTP Applications
Complex ETL

ETL Service No real-time analytics

Security & compliance risks


OLAP DB

Increased costs
OLAP Applications

Copyright © 2024, Oracle and/or its affiliates


MySQL HeatWave overview
Transactions, real-time analytics across data warehouse and data lake, and machine learning in one database service

Social, eCommerce, gaming, healthcare, fintech, IoT apps. Analytics and ML tools.

Queries Results

MySQL HeatWave Object Store

Database
Analytics Autopilot exports
OLTP In-database ML

Data remains in object store,


processing is done in HeatWave
For both non-MySQL and MySQL workloads
Copyright © 2024, Oracle and/or its affiliates
One database is better than two

One service for OTLP & OLAP

No ETL duplication
Transactional
applications
Unmatched performance, at a
fraction of the cost
Analytics tools and InnoDB HeatWave
applications storage analytics Real-time analytics
engine cluster

Improved security

Applications work without changes

1>2 with MySQL HeatWave

Copyright © 2024, Oracle and/or its affiliates


In-Memory hybrid columnar processing

Copyright © 2024, Oracle and/or its affiliates


Massively parallel architecture

• High-fanout partitioning
• Machines & CPU cores can further process partitioned data in parallel
• Optimized for cache size and memory hierarchy of underlying hardware

13 Copyright © 2024, Oracle and/or its affiliates


Overlap compute with communication

Copyright © 2024, Oracle and/or its affiliates


Backups
Manual or Automatic

• Retention Period
• When to Backup
• Full or Incremental
• Point-in-Time Recovery (only non-HA DB Systems)

Copyright © 2024, Oracle and/or its affiliates


High Availability
Single-click HA

Copyright © 2024, Oracle and/or its affiliates


High Availability
Single-click HA

• SLA 99.99%
• Automatic failover
• Zero Data Loss during failure event
• Option to manually switch-over
• Rolling upgrades during maintenance

Copyright © 2024, Oracle and/or its affiliates


Read Replicas
Scale Read Workloads

• Increase capacity for read-intensive workloads


• Add and remove Read Replicas for horizontal elasticity
• Easy to deploy and maintain
• Built-in Load Balancer for the read-only endpoint
• HA Cluster is ready for switchover or failover

Copyright © 2024, Oracle and/or its affiliates


Read Replicas
Single-click Read Replicas

• A single click creates a Read Replica


• Provision
• Launch
• Setup Replication
• Monitor and Manage

• Read Replicas are associated with a DB System


• RO endpoints in the DB System
• Up to 18 max per DB System
• Requires a shape of 4 OCPUs or larger
• CLI, SDK and Terraform support

Copyright © 2024, Oracle and/or its affiliates


Load Balancer
Use Your Replicas Efficiently

When using Read Replicas a Load Balancer Endpoint is automatically provisioned in your DB System.
• Managed by the service
• Materializes as a Read-Only endpoint
• Round robins traffic across Read Replicas
• Manages Read Replica backends automatically

Copyright © 2024, Oracle and/or its affiliates


Massive amount of data stored in files

• Databases are systems of record


• Files are repository for other types of data (e.g IoT, web content, log files)
• Over 80% of the data we generate is in files
• 99.5% of collected data remains unused
• Lack of time, resources, and expertise to process different data formats across different data sources

Social Voice
Devices

Object Store

Events
Sensors

Copyright © 2024, Oracle and/or its affiliates


MySQL HeatWave Lakehouse
Query half a PB data in the object store—in a variety of file formats

• Query data in MySQL, in the object


store, or across both—using standard
SQL syntax

• Up to 500 TB of data—the HeatWave


cluster scales to 512 nodes

• Querying the data in the object store


is as fast as querying the database –
an industry first!

• Scale out data processing in the


object store, data is not copied to the
MySQL Database: for both MySQL
and non-MySQL workloads

Copyright © 2024, Oracle and/or its affiliates


Very simple to query files in the object store
STANDARD SQL syntax generated by MySQL Autopilot, no human required

1. System Setup
 Run MySQL Autopilot on object store to determine cluster size and schema mapping
 Execute DDLs generated by Autopilot

2. Run query across files and tables


 mysql> SELECT count(*) FROM Sensor, SALES WHERE Sensor.degrees > 30 AND Sensor.date =
SALES.date;

Copyright © 2024, Oracle and/or its affiliates


Challenge #2: Organizations want to leverage
ML and generative AI with all their data

Copyright © 2024, Oracle and/or its affiliates


The pain of using AWS services

Amazon Redshift with RDS and ML Modeling

Amazon RDS

RDS to Redshift Pipeline


Utilizing Lambda, Glue, & S3

Redshift ML does NOT provide in-database ML; exports data to SageMaker via Amazon S3

Copyright © 2024, Oracle and/or its affiliates


Machine learning in action with MySQL HeatWave

OLTP Analytics
Applications Tools

Social
Real-time
Real-time ML
analytics on
recommendations
trends
ECommerce

FinTech

SaaS

Copyright © 2024, Oracle and/or its affiliates


Challenge #3: Organizations spend too much
time on manual management tasks

Copyright © 2024, Oracle and/or its affiliates


Manual management tasks consume resources

• On-premises:
• Database management: provisioning, configuration, backup, HA, patching, security & more
• Operating system management: installation, patching, upgrades...
• Infrastructure management: purchase and maintenance of servers, storage
• Data center management: space, power, cooling, disaster recovery & more

• In the cloud with a managed database service:


• Provisioning: right-sizing a database
• Data loading: optimizing load time, memory usage, encoding, data placement
• Query execution: performance tuning, prioritization of queries
• Failure handling: actions to handle an error recovery

Copyright © 2024, Oracle and/or its affiliates


MySQL HeatWave: fully managed database service
100% developed, managed, and supported by Oracle
Automation MySQL HeatWave

High Availability

Read Replicas

Backup

Query Acceleration

Database MySQL AutoPilot

HeatWave AutoML

HeatWave Lakehouse

Security Patch & Upgrade

Provision & Configure

OS Security Patch & Upgrade


OS
OS Installation

Server Hardware Provisioning & Maintenance

Storage Storage Provisioning & Maintenance

Rack & Space


Data Center
Power, HVAC, Networking
Copyright © 2024, Oracle and/or its affiliates
Machine learning-powered automation for MySQL HeatWave
High query performance at scale, higher OLTP throughput, and the best price performance

Copyright © 2024, Oracle and/or its affiliates


Auto Provisioning
Machine learning prediction of memory usage to estimate cluster size
Traditional Provisioning Flow Auto Provisioning

Provision right
Analyze DB schemas/tables cluster size
MySQL
Guess a cluster size

Iterate
HeatWave
Nodes

Adaptive Sampling
Datasets Accuracy in memory prediction 0.01%

Error %
0.1%
TPCH 1024G 98.4%
TPCDS 1024G 96.9%
Cust A 98.3%
Sample Size
Cust B 96.9%
< 0.1% of data scanned for prediction
31 Copyright © 2024, Oracle and/or its affiliates
Auto provisioning with MySQL HeatWave Lakehouse
How to determine the right cluster size required for processing data in object store?

Copyright © 2024, Oracle and/or its affiliates


Auto Shape Prediction
Determines the optimal instance type
Workload

MySQL • New statistics


Server • Intelligent sampling
InnoDB
statistics MySQL Autopilot SQL interface Predictions over time • ML models inside
the server
Query Metrics Machine learning model
• Periodic prediction
Traces/Logs
(future) Data aggregation and sampling

Shape: Shape:
MySQL 2.16 MySQL
Shape 4.32
Core Core Core Core Core Core
Heavy change Light
I/O I/O
InnoDB 16 GB InnoDB 32 GB Memory
Memory

Copyright © 2024, Oracle and/or its affiliates


Same performance with HeatWave for data in DB or in object store
Develop applications with data on object store without any performance impact

120 Query execution time: 10 TB TPC-H

100

80 1.75 minutes
Query time (seconds)

60 1.3 minutes

59 seconds
40
47 seconds

20

14 seconds 14 seconds
0
HeatWave HeatWave Snowflake Amazon Redshift Google Big Query Databricks
Lakehouse

Significantly improves flexibility and reduces costs


Configuration: MySQL HeatWave Lakehouse: 512 nodes; Snowflake: 4X-Large Cluster; Databricks: 3X-Large Cluster; Amazon Redshift: 20-ra3.16xlarge; Google BigQuery: 6400 slots
Benchmark queries are derived from the TPC-H benchmarks, but results are not comparable to published TPC-H benchmark results since these do not comply with the TPC-H specifications.

Copyright © 2024, Oracle and/or its affiliates


Challenge #4: Organizations want to use
multiple clouds

Copyright © 2024, Oracle and/or its affiliates


MySQL HeatWave is optimized for multiple clouds
Maximum flexibility and choice

Optimized for best price performance in each cloud

Copyright © 2024, Oracle and/or its affiliates


MySQL HeatWave on AWS

• MySQL HeatWave runs natively on AWS, optimized for AWS infrastructure

• Data doesn’t leave AWS – saves egress cost, and avoids compliance approvals

• Lowest latency access to MySQL HeatWave

• Tight integration with the AWS ecosystem – S3, CloudWatch, PrivateLink

• Easier migration from other databases (e.g., Amazon Aurora, Redshift, Snowflake)

Combine 5 AWS services into ONE

Copyright © 2024, Oracle and/or its affiliates


Roadmap of capabilities on OCI and AWS
As of March 2024 - https://www.oracle.com/mysql/roadmap

Key capabilities Available on OCI Available on AWS


Fully managed service

OLTP and OLAP in MySQL

Query acceleration for analytics and mixed workloads

Machine learning-powered automation (MySQL Autopilot for HeatWave and OLTP)*

Advanced security*

In-database machine learning (HeatWave AutoML)

Scale-out data management

Interactive query and data management console Coming soon


Performance and workload monitoring from the console Coming soon
Interactive MySQL HeatWave AutoML console Coming soon
Adding HeatWave to any MySQL shape Coming soon
MySQL HeatWave Lakehouse Limited availability

* Auto thread pooling and auto shape prediction in MySQL Autopilot as well as the MySQL HeatWave database firewall will be available soon on OCI.

Copyright © 2024, Oracle and/or its affiliates


MySQL HeatWave on Azure
Connecting to MySQL HeatWave on OCI from Azure VNET

• Familiar Azure-native user


experience

• Automated identity, networking,


and monitoring integration

• Private interconnect and


networking with < 2 ms latency

• Use Microsoft Azure services


with MySQL HeatWave

• Collaborative support

https://www.oracle.com/cloud/azure/oracle-database-for-azure

Copyright © 2024, Oracle and/or its affiliates


Enabling hybrid deployments
OLTP on-premises, OLAP in the cloud

Copyright © 2024, Oracle and/or its affiliates


DB system Inbound and Outbound Replication

Cross-region replication
• DB System to DB System

Live Migrations
• Minimize downtime

Hybrid deployments
• On-premises and other cloud vendors
• Disaster Recovery
• Capacity bursting
• HeatWave for Analytics, ML, Lakehouse

Copyright © 2024, Oracle and/or its affiliates


Available in your data center
OCI Dedicated Region

Self-contained cloud region

MySQL HeatWave and all Oracle public


cloud services in your data center

Public cloud economics and security

Meet data residency and latency


requirements

Copyright © 2024, Oracle and/or its affiliates


Why MySQL HeatWave for new and existing applications?

Simplicity of OLTP, real-time


analytics across DW and data Unmatched performance and Ready for the
lakes, and ML in ONE cloud price- performance distributed cloud
database service
MySQL HeatWave:
Eliminate the cost and  4X faster than Redshift, 10X better price-perf
Deploy on OCI, AWS, Azure
complexity of separate  4X faster than Snowflake, 15X better price-perf
 1,400X faster than Aurora, 2,200X better price-
analytics database, lakehouse, perf Replicate data from OLTP on-
ML, and ETL services premises apps to MySQL
MySQL HeatWave Lakehouse HeatWave for analytics in the
Avoid the latency and security cloud
 11X better price-performance vs Redshift
risks of data movement  15X better price-performance vs Databricks
between data stores  19X better price-performance vs Snowflake Deploy in your data center with
 22X better price-performance vs BigQuery OCI Dedicated Region
MySQL and Amazon Aurora-
based applications work MySQL Autopilot: automation to achieve
without changes high query performance at scale, higher
OLTP throughput, and get the best price-
performance
Copyright © 2024, Oracle and/or its affiliates
Get started with MySQL
HeatWave

Get $300 in credits and try free for 30 days


oracle.com/mysql/free

Migrate to MySQL HeatWave with free expert guidance


Migration program
Request a free MySQL HeatWave workshop
Ask your account manager

Learn more about MySQL HeatWave


oracle.com/mysql
Copyright © 2024, Oracle and/or its affiliates

You might also like