cassandra-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Jonathan Ellis <jbel...@gmail.com>
Subject Re: Ideal configuration for given hardware
Date Tue, 25 May 2010 05:01:50 GMT
yes, I would do raid1 on 2 commitlog disks and raid10 on the 6
remaining for OS + data

On Mon, May 24, 2010 at 2:27 PM, Aaron McCurry <amccurry@gmail.com> wrote:
> Thanks, a lot!  So for RAID 10, is the thought that the node can survive a
> single disk failure and keep going until a normal maintain cycle?  Also are
> you saying that you would configure a single RAID 10 for the whole box?  OS
> included?  I have 8 x 500 Gig drives, so that would leave me with 2T per
> box, which I think is fine.  But I do have one question, in this
> configuration would commit log writing and data directory
> compaction interfere with one another?  Just based on what I read, it seems
> as though you want at least disks/partitions, one for commit log and one for
> data.  Thanks again for the feedback!
> Aaron
>
>
> On Mon, May 24, 2010 at 3:12 PM, Ian Soboroff <isoboroff@gmail.com> wrote:
>>
>> My data disks on two of my nodes are RAID-5, just because of
>> circumstances.  My other nodes are JBOD.  I don't notice any real
>> difference, but I haven't strongly benched it.
>> Ian
>>
>> On Mon, May 24, 2010 at 2:45 PM, Jonathan Ellis <jbellis@gmail.com> wrote:
>>>
>>> I can think of at least 2 clusters running 32GB boxes with single
>>> Cassandra processes on each.  (16 seems to be more common.)  At 64 I
>>> would seriously consider multiple processes per machine.  You'd want
>>> to configure a Snitch such that same-machine boxes were considered the
>>> same rack, there is no separate closeness level of same machine.
>>>
>>> At 32 I think you're fine with one process.  Watch for latency spikes
>>> and see how it goes.
>>>
>>> I would run raid 10 on the data disks if you can afford giving up the
>>> space, otherwise raid0.  I don't know that anyone's tested raid5.
>>>
>>> On Sun, May 23, 2010 at 3:30 PM, Aaron McCurry <amccurry@gmail.com>
>>> wrote:
>>> > I am planning on setting up a Cassandra cluster on a small 16 node
>>> > cluster
>>> > (possibly 32 way).  Each machine has 8 cores 32 Gig of ram and 8 hds.
>>> >  My
>>> > first thought is to setup one of those hds for the commit log, 6 for
>>> > data
>>> > and leave one for the OS.  However I do have a concern about best
>>> > utilizing
>>> > my memory, should I run a larger heap?  Should I run several cassandra
>>> > processes on the same box?
>>> > My concern about the larger heap is because GC's typically get slower.
>>> >  And
>>> > if I run several procs, does cassandra realize that it's the same box
>>> > for
>>> > replication purposes?
>>> > I do have other hd conf options, hardware RAID 0,1,or 5.
>>> > Just looking for some general configuration options as well as some
>>> > real
>>> > world successes with similarly sized hardware.  Thanks!
>>> > Aaron
>>>
>>>
>>>
>>> --
>>> Jonathan Ellis
>>> Project Chair, Apache Cassandra
>>> co-founder of Riptano, the source for professional Cassandra support
>>> http://riptano.com
>>
>
>



-- 
Jonathan Ellis
Project Chair, Apache Cassandra
co-founder of Riptano, the source for professional Cassandra support
http://riptano.com

Mime
View raw message