[TriLUG] Memcached and VMware ESX

John Wheeler jwheeler at etherealfringe.com
Thu Jul 2 16:02:04 EDT 2009


Just had a discussion with IT guy, he informed me that we are  
switching to Xen for the mission critical stuff.
Good news :)



On Jul 2, 2009, at 3:18 PM, John Wheeler wrote:

> 2 Dell 2950's
>
> -John
>
> On Jul 2, 2009, at 2:49 PM, Andrew Ball wrote:
>
>> It probably won't slow things down a whole lot.  We use Xen virtual  
>> machines
>> at work running on Debian for just about everything and are very  
>> happy with
>> it.  I think memcached is particularly easy to do well for either  
>> hypervisor
>> as it's mostly just RAM access and network traffic.
>>
>> What sort of hardware are you using?
>>
>> Peace,
>> Andrew
>>
>>
>> On Thu, Jul 2, 2009 at 12:04 PM, John Wheeler
>> <jwheeler at etherealfringe.com>wrote:
>>
>>> Hey guys,
>>>
>>> I'm re-factoring some caching code for our web application and  
>>> working with
>>> IT to implement our new dedicated memcached servers.
>>> We have two machines with 32Gb a piece.
>>>
>>> The nature of our application is that multiple client communities  
>>> receive
>>> the same kind of service but with completely different data sets  
>>> and search
>>> vectors.
>>> We want to balance the load by distributing the heavy lifting of  
>>> some
>>> search vectors to memcahce.
>>>
>>> We can either install thin OS's on each machine and have two  
>>> memcached
>>> servers, or we can lose 4Gb off the top on each box (to the VMware
>>> hypervisor) and create virtual memcached servers.
>>> We run WMware ESX.
>>>
>>> The key is that this has to be as fast as possible.
>>> I am looking for opinions on whether or not running inside the  
>>> hypervisor
>>> will slow the speed of direct memory access by the memcached  
>>> processes
>>> living inside the virtual machines by any significant amount  
>>> relative to
>>> normal memcached response times.
>>> The benefit we get by virtualizing is the ability to provision  
>>> memcached
>>> server pools for individual client implementations, keeping heavier
>>> application deployments bucketed into their own pools so as not to  
>>> overwhelm
>>> and push out the smaller, less accessed implementation's cache  
>>> entries.
>>>
>>> I am not really concerned about losing the 4Gb to the hypervisor,  
>>> but
>>> should running in a virtualized environment like this slow down our
>>> application requests to the memcached server?
>>>
>>>
>>> John Wheeler
>>> Web Applications Developer
>>> jwheeler at etherealfringe.com
>>> 336-255-8004
>>>
>>> --
>>> TriLUG mailing list        : http://www.trilug.org/mailman/listinfo/trilug
>>> TriLUG FAQ  : http://www.trilug.org/wiki/Frequently_Asked_Questions
>>>
>>
>>
>>
>> -- 
>> =======================
>> Andrew D. Ball
>> 勃安
>> deus caritas est
>> http://www.ibiblio.org/adball/blog/
>> http://filebox.vt.edu/~anball1/
>> --
>> TriLUG mailing list        : http://www.trilug.org/mailman/listinfo/trilug
>> TriLUG FAQ  : http://www.trilug.org/wiki/Frequently_Asked_Questions
>
> --
> TriLUG mailing list        : http://www.trilug.org/mailman/listinfo/trilug
> TriLUG FAQ  : http://www.trilug.org/wiki/Frequently_Asked_Questions




More information about the TriLUG mailing list