Posted by Craig Le Clair on March 24, 2008
Forrester recently surveyed 233 IT decision-makers who have plans to implement or upgrade to at least some part of MOSS 2007 and asked: "Which of the following best describes your organization's time line for implementing or upgrading to Microsoft Office SharePoint Server?". The results? 21% will upgrade immediately and 41% will do so within 6 months.
With this level of adoption the issue of scalability comes up more and more. In one sense you have architectural concerns with any solution that scales horizontally, uses banks of load-balanced Web servers, application servers, and clusters of SQL servers on the back end. Add high availability and you quickly get a complex environment. To Microsoft's credit there is quite a bit available on performance guidelines. But looking through these, and coping with notions of site collections, lists, file arrangements, performance of folder hierarchies versus flat files, and automatic versus manual partitioning, the bottom line seems to be that even on the new 64 bit architecture with 4 screaming Intel processors, and SQL 5 -- the upper limit of the content repository is 500GB.
Assuming allocation for index data and other overhead this really allows less then 20 million objects to be reasonably managed. For enterprise solutions that may ingest 50,000 document a day, this would last about a year. Basically these are not huge numbers. Contrast this –- just for kicks -– with the DocHarbor repository, which was originally built for indexing computer output like customer statements. For statement archiving, ingestion runs done at month-end can be 10 million. Meeting the index management requirement meant building configurable hashing algorithms that can create up to 64K database tables. Anacomp, owner of the DocHarbor platform, claims 1 billion objects managed in their high-end IBM Unix environment.
The clear message from Microsoft is that MOSS 2007 can be an enterprise solution but firms contemplating this should plan carefully. The point for companies adopting SharePoint to consider is the degree of complexity and attendant soft costs required to manage this environment as it grows. I fear that firms will be seduced by the ease of rolling out SharePoint and at the end of the day may prematurely push this platform too far. But let's be fair. All platforms at the "enterprise" level of performance are complex and require strong engineering to maintain, whether it be EMC, IBM, OpenText, Hyland Software, or others. Yet, these products have had years dealing with approaches to keep table sizes manageable, optimize database schemas, index documents at high rates, and maintain business continuity.
Companies that have these mature environments in place must compare the complexity of scaling with SharePoint to leveraging that back end investment and skilled people they have. Many will chose to link that back-end to elements of SharePoint that makes sense.
I'm doing an AIIM Webinar on April 9 that will discuss this. I'd be very interested in any comments from enterprises -- or suppliers -- with experience wrestling with this decision or that have had success or issues with trying to scale SharePoint. Our recent adoption data makes this a very topical issue for many firms.