5
votes

I know there have been similar questions asked over the last few months however looking at a Data Center Operations and know there are some really smart people out there that might be able to help.

Looking for some staffing best practices based on first hand experience and was hoping that there is some experience in this area that can provide "best practice" application:

Three High Availability (99.99% plus) Enterprise Level Data Centers geographically dislocated, one manned 24x7x365, one lights out, one co-location running HOT-HOT-HOT supporting a global community. More than 2,000 operating systems consisting of 95% Windows, 5% Linux and Solaris, 45% virtualized, more than 100TB storage. No desktop support, no Network Administration (administrated separately), running N+1 and serving more than 250 Billion page views annually.

Based on experience what has been your experience with Server to "Data Center System Administrator/Engineer" ratio?

Thanks in advance for your responses.

2
  • As this question cannot possibly have a right or wrong answer I suggest you make it community wiki, or risk having it closed as too subjective. Jun 2, 2010 at 2:11
  • John - thanks for your comment. I understand that there is possibly no right or wrong answer to this depending on your application. I was hoping that some folks with some real world experience in this area could provide their experiences and how they might have applied it to their IT staffing levels.//bob
    – Bob
    Jun 2, 2010 at 14:06

1 Answer 1

2
votes

You provided a lot of data, but I think still more would be needed to give some kind of estimate on that number. First of all, what kind of duplication is present in your system? In other words, do you have 4 sets of 500 things, each running exactly the same server config? Or do you have 200 different servers/services, and only 10 boxes running each thing.

I think you need to go to the base level of each service, determine the activities necessary to support that service, then staff it appropriately. Since you will clearly have some duplication of expertises across multiple services, you may be able to consolidate your head count based on that, but I would start bottom up, rather than top down.

1
  • Zak, Thanks for your response and question. We are running a shared multi-database back-end with 8 middleware applications and a web based front-end. Pool size for middleware varies depending on application, some are as many as 10 servers across all DCs others have more than 100 servers. From a webserver perspective it will be dependent on the traffic for the individual client, no less than 6 servers across multiple DCs ranging to 18 servers. Thanks again...//bob
    – Bob
    Jun 2, 2010 at 14:16