By: Philip Howard, Research Director - Data Management, Bloor Research
Published: 15th April 2013
Copyright Bloor Research © 2013
On April 3rd IBM put out a press release called "IBM Announces New Innovations to Help Organizations Benefit from the Next Natural Resource: Big Data". It discussed the introduction of the IBM PureData System for Hadoop, which includes a new version of InfoSphere BigInsights; the latest version of InfoSphere Streams; enhancements for time series processing in Informix, and something known as BLU Acceleration. However, it forgot to mention the latest release of DB2 (v10.5), which is where BLU acceleration has been implemented (along with Informix). Whoops!
There are some really big deals in this release but the most important is probably BLU Acceleration. This is so significant that I will be devoting a separate article to discussing it and how it works.
As far as the other products are concerned the most important is the IBM PureData System for Hadoop. Like the other PureData models this is an appliance that is pre-balanced with the software pre-installed. It is based on InfoSphere BigInsights, using GPFS storage rather than HDFS. I have discussed this previously but, briefly, it is similar to HDFS but stores data in 128MB blocks rather than 64MB blocks and it has all the sort of robustness and resilience that you would like to have from HDFS but which isn't actually there.
However, this is not all. BigInsights has been extended in a number of ways but the most significant is the introduction of what IBM is calling BigSQL, which allows you to use ANSI standard SQL to address the database; and, also built into the appliance, is data archival capability.
Of course the PureData System for Hadoop will cost you more than doing this all for yourself but where do you want to put your resources: into getting value for your data or in spending (wasting?) time configuring, implementing and maintaining your own DIY solution.
As mentioned, Streams has also been enhanced, although there is nothing revolutionary about this and both Informix and DB2 have had BLU acceleration implemented. DB2 also has a variety of other new and enhanced features, as you would expect. The one thing that hasn't been announced formally is that a JSON storage engine for DB2 is now available as a preview. I discussed this previously in my article questioning whether a relational epithet is still appropriate for DB2 (www.bloorresearch.com/analysis/11839/db2-relational-epithet-longer.html), with JSON to be the fourth storage engine to be available. One thing that hasn't featured in this release, and which I had hoped for, is an inference engine to support DB2's graph store.
Posted: 16th April 2013 | By Philip Nelson :
I'd agree with you that the BLU announcements tended to overshadow the DB2 10.5 announcements, and the rest of the new functionality.
Personally I think that some of the other features have a much wider important than BLU. In particular the ability to build a pureScale / HADR combo giving local scaleability and resilience with remote disaster recovery capabilities is a major step forward.
The JSON support also has wide relevance for eCommerce applications. Since I'm just finishing implemented an application on top of DB2 which makes heavy use of jQuery DataTables it would have been good to have this functionality now (but at least I should be able to move the relational JSON conversions down into the database fairly easily later given that I knew this functionality was coming).
Posted: 17th April 2013 | By Philip Howard :
I am sorry to have to say that IBM has now told me that JSON documents will be stored in tables. This is completely contrary to everything the company has told me over the last 6 months. This is a lot more than a "boo-boo". Not only does it suggest that IBM's left hand doesn't know what its right hand is doing it's also going to disappoint people like Philip above. Moreover, putting JSON into tables is nuts. The best place to put JSON is in a JSON store: I don't think IBM will be able to compete with document databases if it doesn't do this. And if it is going tot try then it would do better to put the JSON documents into its XML storage engine. Converting JSON into tables makes no sense at all.
The messages above were all contributed by IT-Director.com readers. Whilst we take care to remove any posts deemed inappropriate, we can take no responsibility for these comments. If you would like a comment removed please contact our editorial team.
We automatically stop accepting comments 180 days after a post is published. If you would like to know more about this subject, please contact us and we'll try to help.
Published by: electronicdawn Ltd.