* NetForecast report: "Why Centralizing Microsoft Servers Hurts Performance"
It’s no secret to enterprise network managers that centralizing servers in a corporate data center and serving applications to many remote distributed end users causes multiple performance issues. And it’s no secret to vendors that have emerged hand over fist in the past few years to address the inherent problems in application delivery over the WAN.
But despite the myriad technologies available, performance slowdowns are not easily solved, especially not with more bandwidth or faster servers, because the problems often lie in the “chattiness” of the application protocols and the latency that occurs when several roundtrips between servers and clients must be made to fulfill a request. In a recent report from research firm NetForecast, analysts Peter Sevcik and Rebecca Wetzel explore how applications that use Microsoft’s Common Internet File System (CIFS) and its Messaging Applications Programming Interface (MAPI) will suffer performance degradation when served from a centralized data center.
“Despite its many business benefits, server centralization inevitably stretches the distance between users and servers, which can degrade performance to the point that users become disgruntled and unproductive,” the analysts write in the paper. The report entitled “Why Centralizing Microsoft Servers Hurts Performance” gets into great detail as to why performance degrades, and also proposes a few fixes.
“A number of techniques can be brought to bear on CIFS and MAPI traffic to fix poor performance without fixing – or even touching – the adversely affected Microsoft applications, themselves,” the report states.
For instance, NetForecast recommends IT managers deploy technologies to perform the following tasks:
* Reducing turn counts by streamlining application protocols.
* Reducing application payload through content caching and compression.
* Improving network protocol behavior and thus increasing available bandwidth.
* Offloading functions to speed server processing.
To read more specifics on this issue and how to address it, click here.
