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.

I'm currently experiencing a problem whereby load on one guest is adversely affecting load on other guests located on the same host. What I'm curious about is what best practices are there for optimizing a Linux guest.

Typically, the guests are either RHEL 4/5 or SLES 10/11. If they are RHEL, there is a good chance they are part of some oracle monolith such as peoplesoft and filling the role of webserver, appserver or database.

Thanks, Matt

share|improve this question
    
Some details on the ESXi host hardware and overall configuration (CPU, Memory, Network, Storage) would be useful additional information. –  Helvick Dec 15 '10 at 3:06
    
Combination of HP DL380 and BL380's. Typically 4 CPU and up to 32GB of RAM per host. –  Matt Delves Dec 15 '10 at 3:34
    
Also, guest configuration varies a fare bit. Though typically 2CPU guest with 2 or 4 GB of RAM and twice the RAM for SWAP. –  Matt Delves Dec 15 '10 at 3:35
    
Make sure you completely understand the implications of SMP virtual machines (not just in ESXi, but in all hypervisors) and why they may often make the virtual machine run substantially slower than a single vCPU. –  jgoldschrafe Dec 15 '10 at 12:42
add comment

1 Answer

With very few exceptions, rules for optimizing virtual guests are exactly the same as optimizing physical hosts -- profile your application and see where it's spending most of its time, then figure out how to make it run more efficiently. If you can't do anything about the app and you're concerned about QoS, your first step is to look into the best way to implement resource shares/reservations/limits appropriately for your environment -- this is exactly why they exist.

share|improve this answer
    
Do you have any good resources that cover optimizing hosts (physical or virtual)? Any best practices with regard to allocating resource shares? –  Matt Delves Dec 15 '10 at 21:53
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.