Mailing-List: contact user-help@cassandra.apache.org; run by ezmlm
Precedence: bulk
Reply-To: user@cassandra.apache.org
Received-SPF: pass (nike.apache.org: domain of isoboroff@gmail.com designates
 209.85.221.192 as permitted sender)
DomainKey-Signature: a=rsa-sha1; c=nofws;
        d=gmail.com; s=gamma;
        h=mime-version:in-reply-to:references:date:message-id:subject:from:to
         :content-type;
        b=QJH1kSb17XtIhuOUP1yWSrXiKucD80rXa0jr7z6ETZw+sH8vH/1CJ/3PTPfQaW5FKb
         E1TZDu8mBcBkZXVweNnukTFAI4YvxvfysJJMPnnPHOpbbxuGQlhEbrXp+CpTEOW5vHaS
         sQ4dFntFG2g+0FvYjQXiVfTsyRrTw91nwaRQ0=
MIME-Version: 1.0
In-Reply-To: <AANLkTik7sPpP72Ed1Its6yDnjX1pWeZGGrlvKLjcqcQg@mail.gmail.com>
References: <AANLkTimHbNtIoIMDjQxDDO36VoTtoCXK5bHAgkw9pipa@mail.gmail.com>
	 <AANLkTik7sPpP72Ed1Its6yDnjX1pWeZGGrlvKLjcqcQg@mail.gmail.com>
Date: Mon, 24 May 2010 15:12:41 -0400
Message-ID: <AANLkTik4uiykYJGsS-YjoT8FTyqCVKYbw_6qxzLJiCmx@mail.gmail.com>
Subject: Re: Ideal configuration for given hardware
From: Ian Soboroff <isoboroff@gmail.com>
To: user@cassandra.apache.org
Content-Type: multipart/alternative; boundary=00c09f89980e2e401204875bd368

--00c09f89980e2e401204875bd368
Content-Type: text/plain; charset=ISO-8859-1

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
>

--00c09f89980e2e401204875bd368
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

My data disks on two of my nodes are RAID-5, just because of circumstances.=
=A0 My other nodes are JBOD.=A0 I don&#39;t notice any real difference, but=
 I haven&#39;t strongly benched it.<br>Ian<br><br><div class=3D"gmail_quote=
">On Mon, May 24, 2010 at 2:45 PM, Jonathan Ellis <span dir=3D"ltr">&lt;<a =
href=3D"mailto:jbellis@gmail.com">jbellis@gmail.com</a>&gt;</span> wrote:<b=
r>
<blockquote class=3D"gmail_quote" style=3D"margin: 0pt 0pt 0pt 0.8ex; borde=
r-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;">I can think of at=
 least 2 clusters running 32GB boxes with single<br>
Cassandra processes on each. =A0(16 seems to be more common.) =A0At 64 I<br=
>
would seriously consider multiple processes per machine. =A0You&#39;d want<=
br>
to configure a Snitch such that same-machine boxes were considered the<br>
same rack, there is no separate closeness level of same machine.<br>
<br>
At 32 I think you&#39;re fine with one process. =A0Watch for latency spikes=
<br>
and see how it goes.<br>
<br>
I would run raid 10 on the data disks if you can afford giving up the<br>
space, otherwise raid0. =A0I don&#39;t know that anyone&#39;s tested raid5.=
<br>
<div><div></div><div class=3D"h5"><br>
On Sun, May 23, 2010 at 3:30 PM, Aaron McCurry &lt;<a href=3D"mailto:amccur=
ry@gmail.com">amccurry@gmail.com</a>&gt; wrote:<br>
&gt; I am planning on setting up a Cassandra cluster on a small 16 node clu=
ster<br>
&gt; (possibly 32 way). =A0Each machine has 8 cores 32 Gig of ram and 8 hds=
. =A0My<br>
&gt; first thought is to setup one of those hds for the commit log, 6 for d=
ata<br>
&gt; and leave one for the OS. =A0However I do have a concern about best ut=
ilizing<br>
&gt; my memory, should I run a larger heap? =A0Should I run several cassand=
ra<br>
&gt; processes on the same box?<br>
&gt; My concern about the larger heap is because GC&#39;s typically get slo=
wer. =A0And<br>
&gt; if I run several procs, does cassandra realize that it&#39;s the same =
box for<br>
&gt; replication=A0purposes?<br>
&gt; I do have other hd conf options, hardware RAID 0,1,or 5.<br>
&gt; Just looking for some general configuration options as well as some re=
al<br>
&gt; world successes with similarly sized hardware. =A0Thanks!<br>
&gt; Aaron<br>
<br>
<br>
<br>
</div></div><font color=3D"#888888">--<br>
Jonathan Ellis<br>
Project Chair, Apache Cassandra<br>
co-founder of Riptano, the source for professional Cassandra support<br>
<a href=3D"http://riptano.com" target=3D"_blank">http://riptano.com</a><br>
</font></blockquote></div><br>

--00c09f89980e2e401204875bd368--