You go away for a few days and look what happens...

I have been on holidays recently and have just finished catching up with things that happened while I was away. For anyone else who might be interested, here's a quick re-cap...

Mosha was forced to cancel his PASS pre-conference session, but then release an alpha version of his MDX Studio tool and then an update to it

Adrian and Nick both announced they they have been working on not 1 but 2 Rational Guides to PeformancePoint Server books. One on the Planning module and the other on the Analyze and Monitor features (come on, with their BSM book and then the content coming out of thier blogs recently does this really surprise anyone? ) These guys are two smart cookies and if you are working in the PPS area or if you need to get up to speed quickly I expect it would be well worth getting hold of these books when they are released.

The SQL CAT team released another excellent whitepaper, this time on SSAS scale out using SAN Snapshots, and while it will probably not be as widely used as the MDX tuning whitepaper, it's still nice to have some details about how large systems can be architected to provide low latency and high query throughput.

There was a Performance Point launch event last week that seems to have generated the usually confusion between "launch" and "release". I've been caught out by this myself once already, so I am now much more careful, but my personal understanding (Your Mileage May Vary) is that "launch" is a marketing term, it's basically an announcement that a new product is about to be officially released. A "release" is what happens when the final product has been built and the CD/DVDs have started to be manufactured.

One interesting bit that I picked up from one of the PPS reviews is the following snippet relating to SSAS 2008 (emphasis mine)

Q: ... What will be the biggest advances with the release of SQL Server 2008?

AP: I'd say that will be in the area of pure data warehousing, with features like intra-partition parallelism, higher data scales, indexing around partitions and other improvements for scalability, performance and data warehousing workloads. We're going after the 40- and 50-terabyte data warehouses.

Intra-partition parallelism, now that's interesting... Currently the formula engine is entirely single threaded and the Storeage Engine can spin up a separate thread for each partition, so I assume that this means that the Storeage Engine will be able to spin up multiple threads per partition. I am guessing that Multi-threading the Formula engine would be pretty difficult and other improvements like increasing the use of Block mode computation will reduce the need for this anyway.  With the increase in the number of cores per processor, increasing the parallelism of software is going to become more and more important.

Oh and in something completely unrelated, this morning the milk man turned up. We have never had a "milk man" in our neighborhood as long as we have lived there, but we are trying out a new service from Aussie Farmers Direct which supports the local farmers in our state and saves us time by delivering fresh goods to our door step.

 [Update: 26 Sep 2007] Mosha has informed me that I was reading to much into the "Intra-partition parallelism" statement, apparently this is referring to the relational engine and not SSAS, but the good news is that apparently SSAS already does this. I knew the Storage Engine in SSAS was multi-threaded, but I was under the distinct impressions that it used one thread per partition, so at least I have learnt something out of this. :)

Print | posted on Tuesday, September 25, 2007 5:51 PM