HI Jeremy,
>> Im not a big VM fan
I am mixed VM fan, I used to work at a company riding the VM wave and
after I finished there I wrote:
http://www.allinram.info/Virtualisation.html
(a 5 pager that noo ne has ever completely read:) The conclusion was
basically, hard disk's do not virtualise due to disk seek contention
between VMs, so running an app 100% in the cloud will work as long as
you dont need a hard disk (which is a killer for databases) or your
cloud provider supports SSDs (may exist now).
That said, redis w/ virtual-memory turned off runs in a VM like a
champ (because it avoids hard disk seeks)
I am trying to optimise for speed and thruput, (lots of req/s and low
concurrency) and ultimately having 2 redis VMs running on the same
physical host would have each running at close to 50% max req/s, 4
instance -> each runs at 25% max req/s, etc....
ImageMagick as a webserver is analagous, but ImageMagick ops use up
lots of CPU and are minimum several K in terms of request/response,
but the idea of reading up on similar applications performance in VMs
is a good starting point. It is hard to find an application w/ similar
resource usage patterns to redis.
The problem w/ stats on databases running in VMs is disk plays a
dominant role so the stats are useless for non-virtual-memory redis
instances. Maybe some of the in-memory-databases have written up their
findings.
I was just hoping somebody had experimented w/ redis as a VM, there
are a lot of variables to take into account to find the perfect VM
fit.
- Jak
> >
redis-db+u...@googlegroups.com<redis-db%2Bunsubscribe@googlegroups.c om>
> > .