SAS gets close to the database
One of the big announcements at the Teradata user conference this week (confusingly named “Partners”) is SAS integration. Now, SAS is integrating with other MPP data warehouse appliance vendors as well, but it’s likely that the Teradata integration is indeed the most advanced. For example, one customer proofpoint offered was an insurer who used this capability to reevaluate its risk profile at high speed after Hurricane Katrina. I doubt any of the other SAS/DBMS integrations I know of were in customer hands a year ago.
Three still-open questions I hope to address over the next couple of days are:
1. How does this related to Teradata’s own new scoring-on-the-fly capabilities? My best guess is that they’re saying “Do the modeling in SAS while we doing the scoring; SAS is saying “We want to do both modeling and scoring”; and the two vendors will agree to disagree.
2. To what extent does this relate specifically to MPP vs. SMP architecture? I asked that at the public session and got an answer that didn’t really hit the main point, which I think is this: The benefits of doing processing “closer to the data” have a whole lot to do with what the alternative is. If it’s just shipping a result set over the network from an SMP DBMS backend to an application server, that’s (I’m guessing) nice but not huge. After all, you can run an app server and DBMS on the same box if you really want to. On the other hand, if you’re benefitting from Moore’s Law MPP cheapness, that’s (I’m guessing) a whole different matter. Except …
3. Uh, how MPP is SAS right now anyway? Running stats at each node of an MPP system, just on the local node, does NOT give the same results as running them on an integrated result set, absent some fancy footwork. And depending on the join criteria for the extract, it may not even make sense. So I’m confused, and looking forward to doing more research and figuring this out.
Comments
Leave a Reply