Teradata and SAS join forces

Written By:
Content Copyright © 2012 Bloor. All Rights Reserved.
Also posted on: The IM Blog

Over the years SAS has worked with many data warehousing vendors. In particular, in the recent past it has collaborated with a number of database vendors to support in-database analytics. However, as far as I am aware the recent announcement of the “Teradata Appliance for SAS High-Performance Analytics – Model 700” is a first, not just for SAS but for any two companies that are respectively in the warehousing and analytics arenas.

First, what is it? As the name suggests it is an appliance that has been optimised to run SAS analytics by leveraging in-memory processing as an extension to the existing capability of running in-database. The big advantage of that probably doesn’t need rehearsing by now but it means that you don’t have to sample your data and you can run analytical data preparation and model development much faster. The appliance comes in three models with 2, 3 or 4 cabinets, each holding 96 cores, 10.2Tb of uncompressed disk space and 768Gb of memory, with SUSE Linux pre-installed.

One immediate observation is that these are decent enough capacities but aren’t what we would nowadays call “big”. This leads me to suspect that the “Model 700” suffix may only be the first such model to be released, though this perhaps depends on demand. I would expect this to be significant. After all, you have the leading analytics vendor teaming up with the leading pure-play data warehousing vendor.

And this is the point that I want to discuss. What implications does this collaboration have for the warehousing and analytics communities more generally? Of course, one obvious possibility is that Teradata might merge with SAS when Dr Goodnight retires but he has shown no sign of wanting to do so, so I’ll leave that thought in abeyance.  A second point is that this is clearly set out to be a competitor to SAP HANA but that is so obvious that it barely needs mentioning.

More generally you have to bear in mind that Teradata and SAS are in somewhat different positions. While Teradata is a major player in the warehouse space it faces lots of competition from all sorts of other vendors. SAS, however, is the (as opposed to a) major player within the business analytics space. SPSS (part of IBM) is its closest rival but is significantly behind SAS in terms of both market and mind-share. Other suppliers, such as KXEN, are well down the pecking order when it comes to providing competition. There is, of course, R, which is becoming increasingly popular but that’s a (statistical) programming language rather than directly solutions-oriented. Further, SAS provides not just analytics (by which I mean data mining and such-like given this context) but also traditional BI where, of course, it has lots of competition from the likes of Cognos (IBM), SAP Business Objects, Oracle, QlikTech and a host of others. But what SAS doesn’t have is any competitors that offer both BI and analytics based on a single software infrastructure.

So I would have to say that Teradata needs SAS more than SAS needs Teradata, though Teradata brings a big stick to the party in many major accounts. More to the point, however, is that having collaborated in this way, any potential user of SAS, or existing customer that might be interested in migrating to a new database, is bound to look at this appliance as the first port of call. Other things (such as price) being equal why would you look beyond the Model 700 for a solution, provided the processing capacity is right? Normally you would say that performance would be your major concern but it seems unlikely that anybody is going to outperform this particular combination.

All of this raises the question of whether we are going to see more such alliances and/or closer partnerships between vendors. Will any other vendors persuade SAS to come up with similar appliances? My guess would be no. We can expect DB2, Netezza and/or Informix more tightly integrated with SPSS and Oracle has its own data mining capabilities but that doesn’t leave many other analytics/data mining vendors for data warehousing suppliers to partner with. There’s KXEN and StatSoft and if you have open source pretensions you might consider partnering with the open source KNIME (Konstanz Information Miner) but that’s not enough to go round and, in any case, none of these has the market presence of SAS. As a result it looks as if, with the exception of IBM, SAS has the warehousing market by the short and curlies.