IBM DB2 Analytics Accelerator for z/OS V3.1

Yesterday was a prolific announcement day for IBM. Rumours of improvements for the IBM DB2 Analytics Accelerator (IDAA) were already heard before the summer, but now they're officially announced : "IBM DB2 Analytics Accelerator for z/OS V3.1, together with related prerequisite hardware, is a high-performance software offering that delivers fast analytic query responses (ZP12-0503)". We're of course talking about the Netezza technology which replaced the Smart Analytics Optimizer last year.
In short, DB2 tables are offloaded to the Netezza box. When a query is launched DB2 decides whether to fetch the data in its own tables or to go and get them from the Netezza box. The IDAA has proven to be way more performant than queries executed on the mainframe itself. It makes analytical queries possible which were unthinkable of in the past. This happens all transparently to the user or application program launching the queries.

So, what's new with this new version ?

The IDAA can now function as a High-performance storage saver : "this feature stores a DB2 table partition or full table of data solely on the accelerator, reducing the need for host storage. This feature removes the requirement for the data to be replicated on both the DB2 storage and the Accelerator storage. Tables can now be divided between traditional database resident partitions on DB2 for z/OS and high-performance storage saver on the DB2 Analytics Accelerator". The implication is of course that you have data which is only stored in one place. In case the IDAA is unreachable, you don't have access to those data. On the other hand, it could be a great solution to put archived data not only on tape with rather slow access, but also keep them on the IDAA for fast access when needed. In other words : another opportunity to improve and speed up your analytical queries.

Another complaint that was sometimes heard was that customers needed (even for their Data Warehouse) near real-time accessability. This was not possible until now as you always had to copy at least on a partition level. Now we have the incremental update feature. "With the new incremental update capability, tables within the accelerator are continually updated throughout the day. This technology reads the log of the database residing on DB2 for z/OS and applies those updates to the DB2 Analytics Accelerator. With this feature enabled, queries off-loaded to the DB2 Analytics Accelerator operate on a near real-time version of the data". Of course you still control the timing for those updates yourself. Nice improvement !

Also the UNLOAD of the data has been changed via what's called Unload lite. "With UNLOAD Lite, the data conversions are done on the accelerator and not on z/OS". This implies that you can save some MSUs with this.


The DB2 Analytics Accelerator for z/OS software requires :z/OS 1.11, or later,DB2 V10.1 for z/OS, DB2 Utilities Suite for z/OS V10.1, XML Toolkit for z/OS V1.10, IBM Ported Tools for z/OS V1.2.0.

This new version will become available on November 30, 2012.

Short takes on DB2 resources, IBM PureSystems, IDAA, Destination z and IBM Tape turning 60

It's been a while since I had some short takes. Sometimes I'm making a bookmark thinking 'I'm going to blog about this'. But then, it's often too short, too little, just a link, but still, interesting enough to mention. So here we go.

• Susan Visser (also on Twitter as @susvis) writes an excellent blog on DB2 resources. The focus : "Build your Skill on Information Management : books, certifications, tutorials, and more". And here's an very interesting post on Skills Resources for DB2 for z/OS Enthusiasts containing magazines, books, eBooks, Twitter, Redbooks, blogs, events, certification material, communities . . .  

• Just some pictures this one : Roger Luethy with his Storage CH Blog directs us towards a peek inside the new IBM PureApplication System. Go have a look over here and . . . watch your step !

• On a somewhat more serious note is a blog post on Dancing Dinosaur about the relation between the newly announced PureSystems and zEnterprise : 'PureSystems Joins zEnterprise Hybrid Family'. When to use zBX, when to use PureSystems, how are they related, do I not end up with 2 separated hybrid worlds, indeed all very interesting questions that Alan Radding is trying to answer. Just one small quote : "Looking ahead, IBM already is planning zBX support for the next generation z and promises to more tightly integrate the zEnterprise with PureSystems".

• What about IDAA ? Well, Willie Favero blogs about enhancements to the IDAA (IBM DB2 Analytics Accelerator) in this post. The enhancements are due for this summer and they are quite interesting : IDAA will e.g. be updated to allow an incremental update. Now I find that great news since "this also means that the data in the Accelerator can be continuously updated, giving queries an almost real time copy of the data". Also the capacity of the IDAA is raised from 128TB up to 1.280TB. you can find more information over here in an article of Alan Meyer in the IBM Data Management Magazine.

• I mentioned the tips and techniques contest on the Destination z site earlier. Well, here are the finalists. Have a look at them and decide which one is your favourite. Then go over to the voting page where you can cast your vote until next Friday. You should do it, because for the moment it's a head to head race between two of the contestants. You might cast the decisive vote.

• Last but not least we have a special anniversary this month : IBM tape has turned 60 ! Here's a brief history of it. Or you can have a look at several storage blogs or at the IBM site itself. To me the titles up there summarize it very well : "Tape is an essential component of today’s storage infrastructure" and "60 years of innovation – and counting".
  I still know what was the first tape I ever came across in my career, about 25 years ago. You know why : I kept one of those when they disposed of them in favour of some shiny, brand new 3480s. So here it is ! (I'm not even sure it's IBM material. Anyone ?).
  So, happy anniversary to IBM tape !

 

And . . . as I always say, do check them out !

IBM DB2 Analytics Accelerator for z/OS and the DB2Night show

In my last post I mentioned IBM DB2 Analytics Accelerator for z/OS or IDAA (pronounce : Eye-Da) as the successor of the Smart Analytics Optimizer. I didn't want to enter a discussion of why IBM chose to abandon the zBX ship and keep the Netezza hardware boxes. But I just have to repeat my statement I also made in my previous post. Obviously it's a very specific hardware configuration that doesn't seem to fit onto the zBX and there's also very little configuration about it. But you can say the same of every appliance like e.g. the DataPower appliance. And I still think it somewhat undermines the whole zEnterprise message with the centralized role of the mainframe and the management of the distributed appliances on the zBX including Workload Management. I guess (hope) there will be some (financial) regulations for people whe bought a zBX solely for the ISAOPT and who are now moving on to IDAA. I won't even think about the possibility of next generations of IDAA running on zBX.

Let's move on to the real subject of this post : the IBM DB2 Analytics Accelerator (IDAA).

IDAA delivers even faster response times for your Data Warehouse queries with the Netezza technology. The DB2Night show of last Monday was a presentation by Namik Hrle on this very IDAA and the replay is available over here. If you're even only remotely interested in the technology, it's definitely worth an hour's time. Topics discussed are :

• Business and Technology drivers
• Key design and operational features
• DB2 integration explaining every DB2 parameter, Explain functionalities etc.
  
• Powered by Netezza
• Supported Workloads

I won't go into any detail as Namik Hrle does a far better job explaining all the aspects. Just a few quotes to tease you into watching the replay

• You can also use the Explain function when you don't have the IDAA yet as a kind of projection tool
• Loading takes about 0.87TB up to 1.5TB per hour depending on the data
• There's a nice slide summing up the conditions for routing queries to the IDAA
• There's a compression ratio of about 4x
• IBM is looking into refining the updating of the IDAA data based on inserts and/or updates rather than whole partitions or tables now.
  

So, do have a look at the replay. Furthermore you can also find information about IDAA on the IBM page and in the IDAA Bruochure. I particularly like this phrase from the brochure : 'The Forgotten Query'. You can guess what it is : it's that query you tried running optimizing it to the bit and finally deciding there was neither time nor resources to run it. Well, have another look at it now !