Take the 2-minute tour ×
Server Fault is a question and answer site for professional system and network administrators. It's 100% free, no registration required.

Things change when you are delivering to a large enterprise audience. What are the most important things to think about in relation to your server or server farm when scaling to a very large user base?

share|improve this question
add comment

2 Answers

up vote 2 down vote accepted

Best advise is to not scale until you need to. It is a hard enough balance to get right but you don't want to spend thousands buying new hardware that isn't used, while at the same time you don't want your system (assuming single application system) to go down because of load.

Graphing everything using something like cacti is very useful for watching trends. It'll show you how quickly space is getting used, bandwidth is growing, memory is being used etc. If you can monitor response times in here too it will also give a great view of how upgrades go in future.

Make sure that your sys admins talk to the application people. There are many things that application designers can do to reduce load on your cluster/farm. Same goes that there are many things they can do that'll shoot the load up no end. Smart devs should be able to give points of slowness (through code profiling) which will help the admins in deciding where to spend time increasing speed.

share|improve this answer
    
+1, sysadmins and developers NEED to talk –  James Jan 27 '10 at 10:27
add comment

Understanding the relationships between data-centre-space/power-requirements/cooling/available-budget/time-to-deliver - if you understand what of those are available you're half way there. Also being able to communicate these to senior non-techies in a way they understand. The rest's just technology :)

share|improve this answer
    
+1 for a technically correct answer, but it's a bit like saying "all you have to do to cure all known disease is have a fundamental knowledge of matter, human physiology, human physcology, every possible disease vector, and a 100% effective predictive model for mutations. The rest is just technology" =) –  jj33 May 1 '09 at 15:14
add comment

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.