FAST’s Performance Slowdown
Heard something notable at the Butler Group Enterprise Search Strategy Briefing in late November.
A rep from Scotland’s National Health Service talked through a case study of their use of FAST and offered up some … interesting … metrics.
The customer indicated that they were anticipating growing their system from 11 million documents to 18 million documents … but that this growth would require 22 servers. Considering that NHS employes a staff of roughly 150,000, and assuming all these staff run 10 searches a day for a maximum of … say … 16 hours per day, this is roughly 1 query per second.
This means FAST, for this implementation, needs 22 servers to run 1 query per second across 18 million docs. Without going into all the technical detail, this isn’t entirely surprising given FAST’s dependence on a slew of different technologies (which adds to the complexity of their deployment) and their need to distribute to more and more servers as the amount of content that needs to be located, searched and indexed grows (which presents a challenge for companies whose data pools are increasing … i.e. all of them).
Just for the sake of comparison, Exalead customers get 20 queries per second across 20 million docs with only 1 server — less cumbersome, more efficient and greener than the 22 servers described by NHS.
Especially in this time of economic downturn and budget belt-tightening, it’s even more crucial that businesses get the most IT bang for their buck. Make sure you make the right choice for your information access so you can utilize your important data and preserve your corporate resources.
This entry was posted on Thursday, November 27th, 2008 at 4:34 pm and is filed under Events, Tips and tricks. You can follow any responses to this entry through the RSS 2.0 feed.
You can leave a response, or trackback from your own site.
November 28th, 2008 at 1:32 am
The question is how big these documents are. Likely NHS have much bigger documents rather average page in the Internet, since such institutions tend to store documents in multimedia formats with history of modifications.
So,you need to compare performance on the same document base.
November 28th, 2008 at 1:41 am
As well you need use the same servers and provide the same set of features for users.