I’ve had countless emails from readers asking for a technical analysis of what Oracle announced at Openworld 2012 pertaining to the X3 refresh of Exadata Database Machine. I attended the show, fell ill and subsequently had a a lot of work backlog to clear. I will get to this next week and, not surprising to readers of this blog, I’ll take aim on the following words: “Database In-Memory Machine” as they appear in the new marketing nickname for Exadata Database Machine.
Yes, I will blog the matter but would first like to recommend the following excellent blog posts by @flashdba as they relate to “Database In-Memory Machine”:
Note: Part II has one tiny bit of errata as discussed in the comment section of the post. The post speaks of the cache hierarchy of X3 and includes Exadata Storgage Server DRAM in the aggregate. I need to point out that the DRAM in Exadata Storage Server cells is not used for cache. DRAM in the storage servers is used for management (metadata) of Exadata Smart Flash Cache contents, Storage Indexes buffering (storage read buffers, HC decompression output and send buffers, etc). The cache hierarchy of X3 is quite succinctly host DRAM (SGA buffers and Results Cache) and Exadata Smart Flash Cache (the PCI flash devices accessed via SCSI disk driver through the Linux block I/O layer in the cells).
Very Important Background Information
The freshly-updated Oracle Exadata Database Machine X3 has the same fundamental architecture as X2. Additionally, the performance characteristics of the infiniband network used for data flow from storage to database hosts remains at 40Gb QDR (same as X2). Therefore I urge all to spend the time it takes to view my Critical Thinking instructional videos on Exadata architecture. As a hint to my up-coming blog series on the matter, I’ll point out that the supposed “In-Memory” assets in this “new” Exadata X3 In-Memory Database Machine just so happen to be housed in the storage grid–not the database grid. To that end, a good understanding of how data flow works in X2 and X3 Exadata is critical to fully appreciate just what the term “In-Memory” means to Oracle Corporation.
Click here to access the Exadata Critical Thinking instructional videos.


Hi Kevin you mention that the flash cache is not on the database grid do you think it will be beneficial to move it to the database grid would like to know your opinion
Well, generally speaking it’s advantageous to have data as close the processors as possible. However, a cache in a database grid host is not shared and has RAC coherency concerns so what you are looking at in that regard is a read cache of the sort seen with Database Smart Flash Cache (OEL, Sol only) or EMC VFCache (aka lightening).
My main message is to not confuse some storage-level flash cache (ala Exadata Smart Flash Cache) with In-Memory technology. That’s just ridiculous.
Hi Kevin, agreed that those 2 articles by @flashdba were excellent and certainly explain what an IMDB really is and certainly Exadata X3 isn’t one of those!
It will be interesting to see what new features in 12c are specific to Exadata once it is released. Although I don’t believe that Pluggable databases will be Exadata specific, it will be interesting to see how Oracle push them in terms of database consolidation, improved resource utilisation and also how they are licensed….
Agreed. I cannot speak of 12c other than what I’ve seen in OOW slides. I was not in the beta program and have not touched the bits.
Since it would be released until about February 2012 (my prediction) and will an Oracle “dot one” release (e.g., 12.1) I should think nobody will touch it until the “dot two” release is out. That’s my way of saying 12c should not be on anyone’s radar until 2014. But I could be wrong on the matter.
There are a lot of folks just now planning their 9i/10g -> 11gR2 + hardware refresh migrations. I think we should keep our eyes keenly on 11.2. I should think 11.2.0.4 will be extremely solid stuff.
My speculation is that there will not be many Exadata-only 12c features. It wouldn’t suprise me if I’m wrong on that though. Read the tea-leaves. Oracle has dire need to produce lockin-ware.