Oracle AI Database 26ai Released

Announcement of Oracle AI Database 26ai

At Oracle AI World 2025, one of the biggest announcements from my perspective was the launch of Oracle AI Database 26ai, tying into Oracle’s vision for AI-driven data management. The official press release provides full details here, but the highlight came during Juan Loaiza (Executive Vice President, Oracle Database Technologies) keynote, titled: The “AI for Data” Revolution is Here – How to Survive and Thrive. In his session, Juan outlined how this new release integrates advanced AI capabilities directly into the database, architected for mission-critical workloads, and that the AI database includes dozens more major new AI capabilities. I’ve shared my thoughts and key takeaways from his keynote in a dedicated blog post, which you can read here.

Hasan Rizvi, EVP, Database Engineering, Oracle also spoke about Oracle AI Database 26ai in his session “Oracle AI Database Directions: Innovate Faster, Scale Smarter and Minimize Risk”:

Is Oracle AI Database 26ai new release

Interestingly, under the hood, Oracle AI Database 26ai is essentially 23ai, but it has been rebranded as 26ai. Why has Oracle done this? Because Oracle has packed so many new features and enhancements into this release that they consider it a major version upgrade. In fairness, this strategy works in their favour, as you’ll see later on in this blog post 😎.

It’s important to note that Oracle AI Database 26ai is delivered as a Release Update (RU) rather than a full version upgrade. This means an implicit upgrade is not required, making adoption simpler for existing customers. In fact, this RU effectively replaces RU 23.10, as illustrated below:

  • 23.8 – Oracle Database 23ai (April 2025)
  • 23.9 – Oracle Database 23ai (July 2025)
  • 23.26.0 (replacing 23.10) – Oracle AI Database 26ai (October 2025)
  • 23.26.1 – Oracle AI Database 26ai (January 2026)
  • 23.26.2 – Oracle AI Database 26ai (April 2026)

If you’re already on Oracle Database 23ai (blog post here), the transition to Oracle AI Database 26ai is straightforward, it’s simply a matter of applying the latest RU. For those on older versions, Oracle has made the upgrade path equally simple: you can move directly from 19c to Oracle AI Database 26ai, just as you would have upgraded to 23ai.

Oracle AI Database 26ai, will effectively be the replacement long-term support release:

For more info, see Mike Dietrich, Oracle Database Upgrade Product Manager, blog post here.

Where is Oracle AI Database 26ai available

As with the same narrative as Oracle Database 23ai, at the time of announcement, Oracle AI Database 26ai is available on:

  • Oracle Cloud: which includes OCI, Exadata Cloud@Customer, Dedicated Region, Oracle Database @ Azure | GCP | AWS
  • Oracle Engineered Systems: Oracle Exadata, Oracle Database Appliance (ODA)

Will Oracle AI Database 26ai be available on-premises

This is the question on so many minds! Hopefully, you heard it here first: at the UKOUG Discover 2025 conference in Birmingham, UK, Dom Giles, Vice President of Product Management at Oracle, confirmed that on-premises was never off the cards, it was just a matter of when. This reassurance is significant for customers that rely on on-premises infrastructure, and shows Oracle’s continued commitment to hybrid and flexible deployment models.

This effectively debunks the online speculation suggesting that Oracle would abandon on-premises deployments for its latest database release, which caused quite a stir!

Just to be absolutely clear! Dom confirmed that an on-premises version of Oracle AI Database 26ai will be available, but he categorically did not provide a release date! His slide reinforced this point, by masking the date, meaning the timing is still to be announced.

This is purely my own speculation, but based on the naming convention of 26ai, I believe the on-premises release could arrive very soon 😁, potentially as early as January 2026! It seems logical that Linux would be the first on-premises platform to be available, given that it’s the standard for OCI, Engineered Systems, and was the environment offered during the beta programme. Other platforms would likely follow throughout 2026, completing the rollout and conforming to the new name 😎. Do you agree? Let me know your thoughts in the comments.

UPDATE: see next section for confirmed release date for on-premises 🥳

He concluded with Oracle AI Database 26ai is AI and data architected together, very similar to how hardware and software are architected together for Exadata and we know how well that went. Let’s hope the same success happens with this new release, bringing AI to your data!

Update: On-Premises Release Date Confirmed

Only matter of hours later, I saw that it was announced and my speculation of January 2026 for on-premises x86-64 is now confirmed 🥳

See Jenny Tsai-Smith (Senior Vice President, Product Management, Oracle) blog post confirming the release date here.

If you found this blog post useful, please like as well as follow me through my various Social Media avenues available on the sidebar and/or subscribe to this oracle blog via WordPress/e-mail.

Thanks

Zed DBA (Zahid Anwar)

Oracle Exadata X11M Next Generation Hardware Available Everywhere

Oracle today announced Oracle Exadata X11M, labelled as the Next Generation Intelligent Data Architecture which enable Exadata Exascale Intelligent Data Architure for the Cloud Era.

Hardware Changes

In the database server, the CPU moves from AMD EPYC 9J14 to 9J25, whilst still retaining 96 cores, they are up to 25% faster. Same memory sizes but up to 33% faster DRAM.

In the storage server, the flash is now up to 2.2x faster then 10M, with CPU up to 11% faster and DRAM 33% faster.

The amount of flash, memory, and disk remains unchanged from X10M.

Both storage server high capacity and and extreme flash remain unchanged, however the database server eighth rack and storage server eighth rack is now called database server-Z and storage server high capacity-Z.

AI search are now further accelerated on X11M with AI vector search on database severs up to 43% faster with in-memory vector index (HNSW) queries and AI vector search on storage servers up to 55% faster with persistent vector index (IVF) queries.

The I/O latency was already impressive on the X10M at 17us, but Oracle have managed to reduced this now to 14us to give 21% faster XRMEM read latency and up to 43% faster flash OLTP read latency.

Software Changes

Exadata X11M has a new intelligent power efficiency capability that can limit the power consumption of the database server CPUs dynamically to save power when workload demand is low. It can also intelligently turn off unneeded cores to further conserve energy. Couple with hardware changes, this allows for running more databases and workloads or run same workload using less hardware due to the extreme performance enabling efficiently consolidate.

Available Everywhere

Oracle now has Exadata in all leading cloud providers, making Exadata accessible to even more customers.

Couple this with Exascale, which is currently limited to OCI but will in future be on Exadata Cloud@Customer, OCI Dedicated Region, and multi-cloud environments, will make Exadata even more accessible with low entry cost point and pay per use.

For more Info

Please refer to the following links:

If you found this blog post useful, please like as well as follow me through my various Social Media avenues available on the sidebar and/or subscribe to this oracle blog via WordPress/e-mail.

Thanks

Zed DBA (Zahid Anwar)

Oracle Database 23c now Oracle Database 23ai

Announcement

In case you missed it, yesterday 2nd May 2024 at 5pm (GMT) Larry Ellison and Juan Loaiza announced Oracle Database 23ai under the banner of “AI + Converged Data: Oracle’s Strategy for Data Management”. You can see the announcement here:
AI and Converged Data: Oracle’s Strategy for Data Management – YouTube

I was conveniently at the UKOUG Technology Strategy Day (LinkedIn post here), where Scott Hays, Steve Ramsay, Dominic Giles and Killian Lynch plus others were. They all alluded to big announcement that evening but for obvious reasons couldn’t share the details. But in the leading hours, I had seen posts on social media with the “23ai” name, which lead me to believe the rename was going to be part of the announcement 🙂 Post 5pm, was able to speak freely with Dom and Kilian and it totally makes sense, we’re in the era of AI, it’s the latest buzz word, we were due a letter change for some while now! Oracle’s slogan is “Bring #AI algorithms to where your data lives with Oracle Database 23ai”.

Versions over the eras

8i and 9i were for the era of internet, 10g and 11g for the era of grid, and finally 12c, 18c, 19c, 21c and 23c (up to 23.3) for era of cloud. 23ai is actually from 23.4 onwards and the release that will be the mainstream General Available (GA) release.

What’s New

There so much that new, which Oracle have broken down into 3 areas, see below:

AI for Data

  • Oracle AI Vector Search
  • Oracle Exadata System Software 24ai
  • OCI GoldenGate 23ai

Accelerating App Development

  • JSON Relational Unification
  • Graph Relational Unification
  • Free Developer Databases

Mission-Critical Data

  • Oracle Globally Distributed Database with RAFT
  • Oracle True Cache
  • In-Database SQL Firewall

More details of each feature is available here.

Support

Oracle Database 19c as of writing has Extended Support (waived) is till 30th April 2026, so customers have 2 years to move from 19c to 23ai. Latest dates can be found here:
Release Schedule of Current Database Releases (Doc ID 742060.1)

Availability

Oracle Database 23ai is only available on Oracle Exadata Cloud@CustomerOCI Exadata Database Service, OCI Base Database Service and Azure Oracle Database Service. It also available on Always Free Autonomous Database as well for download in the Autonomous Database 23ai Container Image and Oracle Database 23ai Free. For the rest we have to wait for GA which I suspect is very imminent, weeks not months.

Where to get more info

If you found this blog post useful, please like as well as follow me through my various Social Media avenues available on the sidebar and/or subscribe to this oracle blog via WordPress/e-mail.

Thanks

Zed DBA (Zahid Anwar)

Oracle switches to AMD for full Exadata X10M lineup, what else is new?

Chipset switch to AMD

So back in September 2022, I noticed the Oracle Exadata Cloud Infrastructure X9M switched to AMD, where as the Exadata X9M On-premises and Exdata X9M Cloud @ Customer remained on Intel. More details in my blog post here.

It come as no surprise the whole lineup has now switched to AMD! Why? Well putting it simply AMD has beaten Intel on the number of cores for sometime now and with the scalability of Exadata it make senses to have the upper limit cores that AMD offers. Oracle is now using the AMD EPYC 9J14 processors.

Where has the Persistent Memory gone?

Since Exadata X8M, Intel Optane Persistent Memory has existed in the storage cells, bringing an additional layer to the multi-tiered storage architecture. Persistent Memory has the benefit of being order of magnitude faster then Flash but at the fraction of the cost of DDR memory resulting in the Exadata Persistent Memory Data Accelerator.

Simply put, you can’t have AMD chipset with Intel Optane Persistent Memory, so Oracle had to come up with a new solution which they glossed over in the announcement. Which they replaced the Persistent Memory with DDR5 memory and address the memory from the compute nodes using the new feature called Exadata RDMA Memory (XRMEM). This extends the existing Remote Direct Memory Access (RDMA), to create a new shared read accelerator. Interesting Oracle compromised on the writes accelerator that Persistent Memory gave but they would have done the analysis and probably Persistent Memory was largely fronting reads and the penalty to flush to Flash instead of Persistent Memory probably wasn’t that significant. There is now 1.5TB DDR5 DRAM in storage cell compared to 256GB in X9M with 1.5TB Persistent Memory.

What else is new?

The other highlights are:

  • Increased memory configuration of 512GB, 1.5TB, 2.25TB and 3TB DDR5 DRAM in the compute nodes
  • PCIe 4 replaced with PCIe 5, to give 2x 100Gb/sec active-active RoCE network for impressive total throughput of 200Gb/sec
  • Extreme Flash storage server now has the introduction of “capacity optimised” flash drive combined with “performance optimised” flash cards to give impressive increase of 2.4x over X9M
  • 22TB disks instead of 18TB on X9M, giving 22% increase storage

Those that know me, will know I have been predicting:

  • Full switch to AMD chipset for whole Exadata lineup
  • Some sort of replacement for Persistent Memory
  • 22TB hard disk drives

I’m glad to say I was right on all accounts 🙂

For more Info

Please refer to the following links:

If you found this blog post useful, please like as well as follow me through my various Social Media avenues available on the sidebar and/or subscribe to this oracle blog via WordPress/e-mail.

Thanks

Zed DBA (Zahid Anwar)

OGB Appreciation Day : Exadata X8M

What is OGB Appreciation Day?

The Oracle Groundbreakers (OGB) Appreciation Day formally known as OTN Appreciation Day and ODC Appreciation Day, is a great initiative by Tim Hall aka Oracle-Base.com.  Where we take the opportunity to say thanks to the Oracle Community which includes but not limited to ACEs, Java Champions, Ambassadors and all those who have the Groundbreakers spirit #ThanksOGB 🙂

I wonder what will be the name will be next year 😉

More info on Oracle Groundbreakers Community can be found here:
About Oracle Groundbreakers Community

When is it?

This year, it is on Thursday 10th October 2019 and I have to confess I totally missed it and thus my post is a few days out but I didn’t want to do disservice to the spirit of the initiative.

You can see my previous post here:
2017 – ODC Appreciation Day : Oracle Exadata Database Machine
2018 – ODC Appreciation Day : Oracle dcli Utility

You can see a summary of previous years blog posts here:
2016 – OTN Appreciation Day : Summary
2017 – ODC Appreciation Day 2017 : It’s a Wrap (#ThanksODC)
2018 – ODC Appreciation Day 2018 : It’s a Wrap (#ThanksODC)
2019 – OGB Appreciation Day 2019 : It’s a Wrap (#ThanksOGB) – this year

My Contribution : Exadata X8M

When I was at Oracle Open World 2019 a few weeks ago, Larry Ellison (CTO of Oracle) announced the new Exadata X8M:

IMG_6715

Key point being in-memory performance utilising persistence memory and RDMA Network over Converged Ethernet (RoCE), which I will detail later on in this blog post.

Larry also boasted the Exadata X8M storage is 50x faster then AWS and 100x faster then Azure All flash storage:

IMG_6716

Following the announcement I attended another 2 sessions with Juan Loaiza (Executive Vice President, Mission Critical Database Technologies, Oracle) and Kothanda Umamageswaran (Senior Vice President, Exadata Development)/Gavin Parish (Senior Principal Product Manager, Exadata Development), who gave more details on the Exadata X8M:

IMG_6880

The keys changes are:

  1. 100Gb/Sec RoCE internal fabric
  2. 1.5TB Persistent Memory per storage server/cell

IMG_6881

RoCE Networking

IMG_6882

RoCE stand for RDMA (Remote Direct Memory Access) over Converged Ethernet, which initially from the start of Exadata had been over InfiniBand, however Oracle stated Ethernet has caught up and surpassed InfiniBand giving 100Gb/sec throughput as opposed to 40Gb/sec with InfiniBand which is 2.5 times faster:

IMG_6883

IMG_6910

RoCE uses InfiniBand RDMA software on top of Ethernet, so includes all the optimisation and allows for backwards compatibility:

IMG_6911

Also mentioned is the smart network prioritisation which can prioritise critical database messages such as transaction commits, cache fusion over backups, etc using Class of Service:

IMG_6914

An another nice addition is instance failure detected through use of RoCE, because if all 4 ports don’t respond it confirmed server failure and instantly evicted from cluster:

IMG_6915

Persistent Memory

IMG_6884

The Exadata X8M uses Intel Optane DC persistent memory a new silicon technology that capacity, performance and cost is between DRAM and flash:

IMG_6885

In the Exadata X8M, the persistent memory is shared, just as disks and flash are.  So you get all the benefit of aggregated performance, redundancy, etc:

IMG_6887

The benefit of RoCE with persistent memory is the Persistent Memory Data Accelerator, that allows the database to use RDMA instead of I/O bypassing network and IO software, interrupts, context switches:

IMG_6919

Another benefit of persistent memory is the Memory Commit Accelerator, which like Smart Flash Logging, uses persistent memory to further speed up log writes by 8x using oersistent memory as a buffer which is flushed to flash or disk later on:

IMG_6920

Smart capacity management of persistent memory, so primaries on persistent memory and secondary on flash, which is automatically moved to persistent memory when primary is unavailable:

IMG_6921

If Exadata was not fast enough, all this innovation has lead to the “Worlds Fastest Database Machine” with a astonishing 16 million IOPS with less then 19 microseconds:

IMG_6909

For more information on Exadata X8M can be found here.

Finally Happy OGB Appreciation Day! #ThanksOGB #ThanksODC #ThanksOTN 🙂

If you found this blog post useful, please like as well as follow me through my various Social Media avenues available on the sidebar and/or subscribe to this oracle blog via WordPress/e-mail.

Thanks

Zed DBA (Zahid Anwar)