DB Sizes Perf and Scalability Part 2
I was contacting Kerrie yesterday with the same questions. This is her answer to why their figures differ from Satya’s.
“Hi Stefan, yes we did look at that while we were working on the book. Satya came out with his several months after ours. We take a different approach – he is basing it on the number of events collected per day (ground up). Ours took a look at the growth of the databases over time, in multiple environments. More of a top-down approach. We also include a fudge factor ..
our calculator has been verified in numerous installations, including some by MCS consultants. “
I thought it was important enough to turn into a full post to bring it to peoples attention as not everyone reads the comments.
The big thing is that Satya’s figures are in the Performance and Scalability white paper so will be used by people as “official”. Only people that look at the blogs will know differently. If these are wrong and the disk space needed is significantly larger then some organisations will not be happy. The Product Group really need to come out with an official sizing guide that is credible. It is pretty easy to work out the number of servers needed but organisations get more upset over disk space sizing – especially when you are asking them for several terabytes of space.
I stand by my answer from yesterday to use the figures from the spreadsheet based on Kerrie and Cameron’s work as basing it on real world database sizes is more realistic in my opinion. Plus customers get less annoyed if you overspec and they find out that the usage is not as big rather than underspec and run out of space.