Exadata System Software 24ai – Delivers mission critical AI at any scale

Text Size 100%:

 

The Exadata Product Management and Engineering teams are excited to announce the release of Exadata System Software 24ai. Exadata System Software 24ai, version 24.1.0.0.0, is the latest and one of the most significant releases of Exadata software since the very first version back in 2008. Exadata System Software 24ai is a critical release. It includes powerful new optimizations to accelerate Oracle Database 23ai – including AI Smart Scan, Exadata Live Update, Columnar Cache on Exadata RDMA Memory (XRMEM), new Flash Cache observability, increased number of VMs per database server, and many more.

Exadata System Software 24ai accelerates AI Vector Search by up to orders of magnitude and delivers mission critical AI at any scale. It enables applications to run AI Vector Search for large volumes of data and high numbers of concurrent users, making Exadata the ideal platform for running cutting-edge Generative AI applications.

Below is an overview of the latest and enhanced features included in Exadata System Software 24ai.

AI Smart Scan

AI Smart Scan greatly accelerates Oracle Database 23ai AI Vector Search with optimizations that deliver extremely low-latency, high-throughput, parallelized scans across massive volumes of vector data. AI Smart Scan processes vector data at memory speed, leveraging ultra-fast Exadata RDMA Memory (XRMEM) and Exadata Smart Flash Cache in the Exadata storage servers, and performs highly scalable vector distance computations and top-K filtering where the data lives, avoiding unnecessary network data transfer and database server processing.

AI Smart Scan Overview slide showing transparent offload of Vector AI search to Exadata Storage

 

 

Oracle Database 23ai

 

In-Memory Columnar Speed JSON Queries

In-Memory Columnar Speed JSON Queries comparison chart
In-Memory Columnar Speed JSON Queries

In Exadata System Software 24ai, Oracle Database 23ai’s native JSON data type and 19c’s OSON format documents are now automatically and transparently transformed into the In-Memory Columnar format in Flash Cache and XRMEM. Storing these data in columnar format allows Smart Scan to benefit from SIMD processing, data pruning, dictionary encoding, and compression to accelerate JSON queries up to 6.5x faster.

 

Exadata Transparent Cross-Tier Scan results in up to 3x faster queries
Transparent Cross-Tier Scan

Transparent Cross-Tier Scan

With Database 23ai, the optimizer can now transparently scan and use columnar data from the database and storage tiers in a Cross-Tier Scan, accelerating serial and parallel queries by up to 3x. Transparent Cross-Tier Scan enables a single query to leverage the features of In-Memory scans – data pruning, dictionary encoding, compression, and SIMD Processing – and efficiently utilize the resources of both database and storage servers.

Faster Pluggable Database Startup

Improvements in the parallelization of internal operations while caching the database control file in Exadata RDMA Memory accelerate the startup speed of PDBs up to 2x faster than previous database releases.

Faster Hash Joins of Large Tables

Offload of larger Bloom filters to Exadata Storage Servers results in up to 4x faster analytic query performance
Faster Hash Joins of Large Tables

Beginning with Exadata System Software 24ai, Oracle Database 23ai can offload larger Bloom filters to the storage servers, resulting in up to 2x faster analytics query performance.

Pipelined Log Writes

When deployed on Exadata X10M, Oracle Database 23ai, and 19c (19.22), pipelining redo writes rather than the traditional “batched” redo write mechanism increases transaction performance by up to 1.45x.

Smart Scan on AES-XTS Encrypted Data

Oracle Database 23ai adds AES-XTS encrypted tablespaces for increased security and performance. Exadata System Software 24ai natively and transparently scans tablespaces encrypted with AES-XTS.

Smart Scan during TDE Encryption and Rekey

Smart Scan can scan data in tablespaces while encrypted online or rekeyed, resulting in up to 4x higher query performance.

Smart Scan on Index-Organized Tables

Smart Scan can scan new Index-Organized Tables, including compressed IOTs.

Smart Scan on Wide Tables

Smart Scan can scan new Oracle Database 23ai wide tables with up to 4096 columns.

Smart Scan during Timezone Upgrades

Beginning with Database 23ai, Smart Scan is active during time zone updates for all tables except those with TIMESTAMP WITH TIMEZONE columns.

Smart Scan of Oracle Database 23ai Functions and Operators

Smart Scan includes new date, interval, and Boolean functions and operators introduced in Database 23ai:

  • Date – CEIL and FLOOR functions
  • Interval – CEIL, FLOOR, ROUND and TRUNC functions
  • Boolean – TO_BOOLEAN function, IS FALSE, IS NOT FALSE, IS TRUE, IS NOT TRUE predicate operators

Exadata Smart Scan for wide tables, time zone upgrades and new operators and functions in Oracle Database 23ai

Caching

Columnar Smart Scan at Memory Speed depiction
Columnar Smart Scan at Memory Speed

Columnar Smart Scan at Memory Speed

New in Exadata System Software 24ai, the hottest (most frequently used) columnar data are automatically loaded into Exadata RDMA Memory (XRMEM), enabling Smart Scan to take advantage of high-speed memory and increase performance.

Exadata Cache Observability

Exadata Cache Statistics (ecstat) is a statistics sampling tool that uses the same style as other Linux utilities, such as iostat. It gives customers insight into the current workload on the storage servers, exposing valuable data, including small and large reads per second, throughput, cache hits and misses, and much more.

Automatic KEEP Object Load into Exadata Flash Cache

Adding the KEEP attribute to a table automatically instructs the storage servers to load the object into the Flash Cache without requiring the explicit full table scan. Automatically loading data into Flash Cache improves network efficiency and avoids database server CPU usage by not returning unnecessary query results to the database servers.

Infrastructure

Increased Number of Virtual Machines on X10M

Beginning with Exadata System Software 24ai on X10M and subject to available system resources, the maximum number of VMs per database server is increased to 50, enabling customers to increase consolidation density and further leverage the hardware capabilities of Exadata X10M.

Secure Boot for KVM Virtual Machines

Beginning in Exadata System Software 24ai, customers deploying new VMs can enable Secure Boot for Virtual Machines to increase the security posture of their database environments.

High Availability

Improved RoCE Network Resilience

Exadata System Software 24ai includes automated RoCE network links and traffic monitoring. If a network link exhibits a traffic stall, the Exadata software will move the affected IP address to a partner link, avoiding network-related database and cluster failures. When the stalled link resumes functioning, the IP address automatically returns to the working link.

Exadata System Software 24ai Infrastructure Updates

Monitoring and Management

Enhanced RoCE Network Discovery

Exadata System Software 24ai includes two new utilities for easier RoCE network discovery:

  • configure_roce_hostinfo.sh – discovers and documents the current RoCE switch port configuration for easy reconstruction if needed and as a helpful map for the rocelinkinfo utility
  • rocelinkinfo – provides the real-time status of each link in the interconnect fabric, simplifying management of the RoCE network

AWR and SQL Monitor

X10M Extreme Flash Storage Servers

In Oracle Database 23ai and 19c, X10M Extreme Flash storage servers better differentiate between performance-optimized and capacity-optimized flash in the Automatic Workload Repository (AWR) report.

Automatic Workload Repository (AWR)

Oracle Database 23ai includes the following sections:

  • Top PDBs by Database IO Requests – Showing I/O request details by PDB for the top 50 PDBs in a CDB.
  • Columnar Cache – Previously included in the Flash Cache section, this new section includes detailed Columnar Cache on XRMEM data and the existing Columnar Cache on Flash Cache data

Snippet of Top PDBs by Database IO Requests in AWR

The following sections are improved:

  • Top IO Reasons by Requests – This section also details why I/O was performed on storage servers and now includes the media type (flash/disk) and I/O type (reads/writes).
  • Smart IO – Details added for storage server reads Eligible for Smart IO, Columnar Cache Reads, Columnar Cache Eligible, and Columnar Cache Saved. This has also been included in Oracle Database 19c.
  • Wait Event Histogram – This includes the following wait events and are also included in Oracle Database 19c:
    • cell single block physical read: RDMA
    • cell single block physical read: flash cache
    • cell single block physical read: xrmem

 

Snippet of Top IO Reasons by Requests in AWR

SQL Monitor

SQL Monitor includes the following metrics:

  • Eligible bytes for smart IO
  • In-Memory Columnar format
    • In-Memory Capacity (IM Capacity bytes)
    • In-Memory Query (IM Query bytes)
    • No Memcompress (No memcompress bytes)
  • XRMEM Columnar Cache bytes
  • Bytes added to Storage Index (Storage Index creation)
  • IORM IO requests on Flash and Disk
  • IORM wait times on Flash and Disk

JSON API for Management Server

CellCLI, DBMCLI, and ExaCLI users can now configure these utilities to output in JSON format. In addition, the REST interfaces on the database servers (hosts and guests) and storage servers have been updated to provide JSON output.

Security

SNMP Security

In Exadata System Software 24ai, SHA2 authentication has been added to the existing SNMPv3 available in the Exadata servers to enhance overall system security.

Hardware, Grid Infrastructure, and Database Minimum Requirements

Hardware

Exadata System Software 24ai is available for all X6 and newer Exadata Database Machines and Storage Expansion Racks.

Grid Infrastructure and Database

Exadata System Software 24ai has the same minimum software requirements for Oracle Database and Grid Infrastructure as Exadata System Software 23.1.

Exadata System Software

The minimum Oracle Exadata System Software version required on an existing system before updating to release 24.1.0 is Oracle Exadata System Software 21.2.10 (March 2022).

For details, refer to My Oracle Support Note 888828.1.

Further Information

The What’s New in Oracle Exadata Database Machine chapter of the System Overview for Exadata Database Machine provides more information on the Exadata System Software 24ai release. Documentation for Oracle Exadata Database Machine is available here.

Delivering Mission Critical AI at Any Scale

Exadata System Software 24ai derives its name from exciting new AI Vector Search capabilities in Oracle Database 23ai. However, Exadata System Software 24ai is more than its AI capabilities. With unique new features including Transparent Cross-Tier Scan, Columnar Smart Scan at Memory Speed, In-Memory Columnar Speed JSON Queries, Pipelined Log Writes, a host of new Smart Scan optimizations, Exadata Live Update, Secure Boot for KVM Virtual Machines, Improved RoCE Network Discovery and Resilience, Increased Number of Virtual Machines on X10M and many more, Exadata System Software 24ai is ready to accelerate our customers applications further, enable greater consolidation, increased security and even higher availability. 

Since its inception, Exadata has been the ideal platform to run any Oracle Database workload. Exadata System Software 24ai is an extraordinary release – born from a relentless focus and drive to engineer software and hardware together, continually pushing the envelope of performance, availability, scalability, and security. It increases the performance of existing Oracle Database workloads and propels our customers into the Artificial Intelligence era with smart offload capabilities found nowhere else in the industry. With optimizations that deliver extremely low-latency, high-throughput, parallelized scans across massive volumes of vector data and the proven ability to power extremely high user communities, Oracle Database 23ai, Exadata, and Exadata System Software 24ai are ready to deliver mission-critical AI at any scale.

And this is just the beginning!!

Finally, and on behalf of the Exadata PM Team, I’d like to recognize the phenomenal efforts of the combined engineering, testing, MAA, documentation, and the many other teams that make a release like Exadata System Software 24ai possible. Their relentless focus on quality and exceeding our customer’s expectations is nothing short of awe-inspiring. Thank you one and all. 

Alex Blyth

Senior Principal Product Manager – Exadata

Alex Blyth is a Product Manager for Oracle Exadata with over 25 years of IT experience mainly focused on Oracle Database, Engineered Systems, manageability tools such as Enterprise Manager and most recently Cloud. Prior to joining the product management team, Alex was a member of the Australia/New Zealand Oracle Presales community and before that a customer of Oracle’s at a Financial Services organisation.

Seth Miller

Principal Product Manager

Seth Miller is a Product Manager for Oracle Exadata and has been working with Oracle products at various companies since 2005. Seth has a strong technical background with extensive experience with Oracle Database and Exadata as a DBA, Consultant, Systems Integrator, Oracle University Instructor, and Product Manager. Seth has also been heavily involved with the Oracle user community, previously serving on the boards of two Oracle Users Groups.

Maruti Sharma

Senior Principal Product Manager

Maruti Sharma is a Senior Principal product manager for mission critical database systems at Oracle with over 25 years of software experience focused on relational databases, Big Data, NoSQL data stores, server programming, microservices. Prior to joining Oracle, Maruti was a Chief Architect and Associate Technical Fellow at The Boeing Company where he was responsible for managing everything data related.

Christian Craft

Senior Director, Product Management

Chris Craft is Senior Director of Product Management, focused on advanced Oracle database technologies including Oracle Converged Database, Exadata, Exadata Cloud Services, Autonomous Database, and Oracle’s Zero Data Loss Recovery Appliance.  Mr. Craft has served in multiple roles within Oracle including On-Site Customer Support, Service Delivery Management, Sales Consulting, and Product Management.  His experience ranges from application development, data modeling, and performance benchmarking to managing customer relations, and leading teams of technical professionals.  Mr. Craft has worked with over 500 Oracle customers across virtually all industries and including many of the largest customers in the world.

Carlos Ortiz

Sr. Principal Product Manager

Carlos A. Ortiz is a Sr. Principal Product Manager in the Oracle Database group, focused on Oracle Engineered Systems, including Exadata Database Machine and Oracle Database Appliance. Carlos has 20+ years of product management experience in the Information Technology industry, managing mission-critical systems solutions for databases and applications.