M-A-O-L

This just came in: Teradata acquiring Aster Data. Database consolidation wars is full speed ahead, last month, it was HP acquiring Vertica, last year IBM bought Netezza, and EMC bought Greenplum. So within six months, the four biggest and most promising MPP DB vendors have found a new owner.

Curt Monash has been trying to define Machine Generated Data (but Daniel Abadi doesn’t fully agree) because machine generated data is what’ll be fuelling a lot of the future growth of DB systems. Understanding MGD and its growth pattern will help design next gen DBMS.

Funny enough a recent Economist piece, It’s a smart world, I stumbled upon it via a Teradata blog, is looking at that from the other angle – how human and MGD have to interact to support further growth, using China’s exploding population as an example.

Teradata Partners with RainStor for Big Data Retention and Retrieval. They play it as a Retention thing (which is the right thing to do), but I wonder what that means for the Teradata Extreme Data Appliance, if anything? But even more so I wonder what that means for Rainstor, since this is their first partner deal with a DBMS vendor. Are their more deals in the pipeline, or may this lead to closer ties with Teradata?

Aster Data Founder Mayank Bawa is engaging in an analytics pissing contest with Vertica in Sessionize with Style? …Or How a column-only, SQL-only database and lack of MapReduce, will cramp your style!, all that about a mini-series of Vertica blogs, where a Vertica engineer was pointing out perceived flaws of Aster and Teradata. Let’s sit back and watch!

A quicky: eBay followup — Greenplum out, Teradata > 10 petabytes, Hadoop has some value, and more. Interesting to see that the impression is that Greenplum got thrown out more for reliability reasons than performance. EBay also was repeatedly mentioned as a key customer using the MapReduce integration piece in the past, there’s also an update on that.

Does anybody remember as far back as two months ago? That’s when I asked

All these connectors being announced makes me think there’s somebody out there with a matrix of RDBMS and NoSQL systems, looking at which combinations don’t have a marketable connector yet so he can be first to market.

Now we have another one, according to ZDNet’s Teradata, Cloudera team up on Hadoop data warehousing story, except by now they may be the last to market instead of first… but giving credit where it’s due, MapReduce has been part of Teradata’s strategy for a while, so I guess the new thing is that

The two techniques technologies will co-exist. Teradata will bundle a connector (the Teradata Hadoop Connector) to its systems with Cloudera Enterprise at no additional cost. Cloudera will provide support for the connector as part of its enterprise subscription. The two parties will also jointly market the connector.

Found via myNoSQL.

There are some funny little wars being fought in the blogosphere between Teradata and Netezza attacking Oracle Exadata. Instead of taking side, let me just link to what’s been going on so far.

• The Teradata whitepaper that got the ball rolling: Exadata – the Sequel. Exadata V2 is Still Oracle
• DB Professor Daniel Abadi is defending Oracle Exadata
• Netezza releases an eBook Oracle Exadata and Netezza TwinFin™ Compared
• Oracle’s Greg Rahn responds with Oracle Exadata and Netezza TwinFin Compared – An Engineer’s Analysis
• Netezza’s Mike Kearney replies to Greg Rahn
• David Birmingham lights the fire again, and publishes another article on his Netezza blog: Exadata’s SOA (Secondhand-On-Arrival)

That’s where we are right now. If you can only read two links of the six, then read the first two ones, because they add most value, IMHO.

I’ve had Teradata on the phone yesterday, maybe will talk about that later. For now, here’s what Curt Monash has to say about what he heard from them: Teradata’s future product strategy.

• Single DBMS, capable of meeting all analytic needs while running in a single instance, usually running on …
• … proprietary hardware …
• … built from conservatively-chosen parts.

His update was all triggered by Teradata’s acquisition of Kickfire, which I entirely forgot to blog about.

Stephen Jannise of Distribution Software Advice put together a nice speculate post about Oracle’s next acquisition targets based on an analysis of the last five years worth of acquisitions: Oracle Mergers & Acquisitions: Who’s Next? Teradata, Informatica and Tibco are his straightforward ideas – go read his article for more.

Ramon Chen’s post Simplification Through Specialization (later repeated on the Rainstor blog) rang quite a bell with me, also on the personal level. But here’s why specialization is better for some:

If you were going to spend big bucks, you’d expect results and performance to match your use cases and requirements. With Big Data upon us and years of data warehousing experience, we already know that transactional RDBMS’ are not the best repositories for analytics. Furthermore Teradata, Netezza and a cadre of  new providers such as Cloudera (with Hadoop), Aster Data, Greenplum, Paraccel and others are further specializing in big data analytics & insights.

Similarly, Oracle and other RDBMS’ with all their glorious complexity should be the least likely repository to keep massive amounts of data if long term retention is your focus. Would you use your Ferrari to store your furniture? Features such as two phase commit, referential integrityand even update would be considered over kill for immutable retention.

More on specialization vs. general purpose over the next few days, so check back tomorrow.