Written By:
Content Copyright © 2009 Bloor. All Rights Reserved.
Also posted on: Accessibility

The Madison was, of course, a dance. And it’s where all those bridges are. But it’s also the codename that Microsoft is using for its development of the technology it acquired when it bought DATAllegro. As Madison has now gone into closed technology preview at 10 customer sites (which is an impressive number given the product’s history) and as it is due to be formally released during the first half of 2010, it is worth taking a closer look at what Madison involves, even though details are somewhat scanty at present.

Of course, the big issue was that DATAllegro was based on Ingres and Linux whereas Madison is based on SQL Server and Windows Server. However, it was the approach taken by DATAllegro that was really important: provided that you could support the multi-level and replicated partitioning that was inherent in the original offering then porting the product shouldn’t have been too difficult. Whether that has been the case I can’t say because Microsoft is not letting out the details. On the other hand it is using terms such as “Landing Zone”, which aficionados of DATAllegro will remember, so this suggests that the port has indeed been relatively smooth.

The big issue is not, however, in the technical details, it’s in what Madison will bring to Microsoft. And the answer to that, primarily, is scale. I am sure that performance is also on the radar but it is really scale that is important. Microsoft hasn’t historically been able to compete for the largest data marts or warehouses in the 100s of terabytes. I understand that the company’s largest warehousing customer is around 65TB but even Microsoft itself will accept that its current sweet spot is less than 32Tb. Thus, for example, this is the limit to which SQL Server Fast Track Data Warehouse reference configurations have been tested. And on the assumption that you want to allow for 50% growth in data volumes then somewhere in the low twenties looks a sensible maximum. But Madison should support hundreds of terabytes and, according to Microsoft, petabytes.

In terms of marketing, perhaps the most obvious function of Madison is in reassuring any existing or future Microsoft warehousing users that they will be able to scale to the levels supported by Madison when they need to. This will also apply to users of SAP BW of course, both those who currently use SQL Server and those who might be thinking of doing so. Now, you might guess that this would be the initial focus and that only once this position is solidified would Microsoft seriously seek to win new non-Microsoft warehousing business. However, I am assured that this is not the case: that Microsoft will be targeting new user accounts from the outset. Indeed, according to Microsoft “we are seeing real interest from our competitors’ customers who already see the benefit of migrating to Madison”. Of course, they would say that but it doesn’t mean it’s not true.

It’s too early to go into a lot of detail about Madison but it is coming and, provided Microsoft doesn’t mess up, it will have a big impact on the market.