On Fri, Apr 1, 2011 at 11:00 AM, Mark Steele wrote:
> I've done some rather disappointing tests with Riak using Rackspace cloud
> servers. Much better off on dedicated hardware if you can find it.
I haven't found anything except small web sites that RS Cloud has
enough performance for.
-J
_
>
> If you want good predictable performance but you are happy to live with the
> risk of loosing some of your data ( in the event of a cluster failure where
> the number of nodes that fail > than your n_val ) then run with the local
> ephemeral storage in RAID 5 or 10 and take snapshots of the da
I concur with Mark, Rackspace has some poor performance with Riak.
On Fri, Apr 1, 2011 at 1:00 PM, Mark Steele wrote:
> I've done some rather disappointing tests with Riak using Rackspace cloud
> servers. Much better off on dedicated hardware if you can find it.
> Mark Steele
> Bering Media Inc.
I've done some rather disappointing tests with Riak using Rackspace cloud
servers. Much better off on dedicated hardware if you can find it.
Mark Steele
Bering Media Inc.
On Fri, Apr 1, 2011 at 11:51 AM, David Dawson wrote:
> Mathias and Alexander,
>
>Thanks for both of your replies the
Hi David,
If you want both good performance and reliability you might look at somebody
like SoftLayer who will let you mix and match cloud and dedicated servers to
meet your needs. Both they and Voxel have cloud servers that beat the pants off
EC2.
-J
Sent via iPhone
Is your e-mail Premiere
Mathias and Alexander,
Thanks for both of your replies they were very informative and really
have helped me to make my mind up, but to summarise:
- If you want good predictable performance but you are happy to live
with the risk of loosing some of your data ( in the event of a c
Hi David,
Alexander already gave you a good rundown on EC2 and Riak, but let me add some
of my own experiences running databases on EC2 in general.
The short answer is, Riak is certainly successfully used in production on EC2,
so nothing should hold you back from testing a setup on EC2. But th
All things being equal (which is atarded[0] because I don't know squat
about your use case), I would go with choice #2. Main concern would be multiple
failures before read repair could be executed via #1. Depending on data size
this could take a long time.
Summary:
1. Potential data loss d
