Organizations are using Microsoft® Office SharePoint® Server 2007 to deploy intelligent portals that provide connectivity between people, processes, customers, and intellectual capital. Specific examples include business process solutions such as team collaboration, data records management, document management and publishing, and dynamic score-carding. The growth rate is increasing, with adoption of Office SharePoint Server 2007 which provides increased functionality at somewhat higher resource costs (server CPU, memory, network, etc.) than SharePoint 2003. Many organizations who are moving to the new version are also taking the opportunity to consolidate their distributed deployments and SharePoint farms into fewer centralized data centers. This approach has many proven cost benefits, but requires these new data centers be able to provide consistent, predictable, and adequate service to users who are now located more remotely from the SharePoint farms across a corporate Wide Area Network (WAN).
In distributed farm deployments, server capacity (usually the Web front-end (WFE) servers) was likely the limiting resource to available farm capacity, with local users being connected over the corporate Local Area Network (LAN) at backbone bandwidths of 100Mbit/sec or even 1Gbit/sec. By moving the farms to a centralized location, potential limits to capacity and performance based on available WAN bandwidth (for example, a T3 connection at 44Mbit/sec, or even a T1 connection at 1536Kbit/sec) can be created. Further, the introduction of latency (delay) due mainly to connection distance, and some small percentage of dropped packets will have an impact on performance.
Why has the network now become more important?
One approach to improving overall system performance is the use of a class of Traffic Management Device commonly referred to as a WAN Accelerator. This paper details the results of using one such solution - the F5 Networks® BIG-IP® Local Traffic Manager (LTM). By enabling the WebAccelerator module on BIG-IP LTM, test results showed a significant increase in user-perceived performance while at the same time decreasing the load on the WFEs. This technology was tested in the HP SAE Nashua Labs in two typical customer scenarios. A "branch office" scenario representing corporate users accessing SharePoint performing read/write operations; and an "Internet" scenario representing anonymous users accessing (read only) a company's externally-facing Internet portal. The two solution deployments, best-practice configurations and observed performance results are detailed in the following sections; and demonstrate how significant performance improvements can be realized.