By Girish Chhabra (1) on June 2nd, 2010
Recently, while working on a SQL server optimization project, we had the opportunity to look into one interesting problem. We had a huge database (to the tune of 800GB) which was being hammered with approximately 30000 transactions per second. Database load was expected to grow by a factor of 100 in coming days and the idea was to devise a solution which could handle that load. This was a SQL server 2005 enterprise edition database hosted on an 8 processor fifth Generation server. We wouldn’t say this server was on its knees with this load but yes there were wait times longer than expected and to add to that there were times when data traffic suddenly went up significantly and in those times DB was not able to keep up.
