Mailing-List: contact cassandra-user-help@incubator.apache.org; run by ezmlm
Precedence: bulk
Reply-To: cassandra-user@incubator.apache.org
Received-SPF: pass (nike.apache.org: domain of richiesgr@gmail.com designates
 216.239.58.184 as permitted sender)
DomainKey-Signature: a=rsa-sha1; c=nofws;
        d=gmail.com; s=gamma;
        h=mime-version:date:message-id:subject:from:to:content-type;
        b=YYQx0C6koSo3jv9Qx4yjkMjdW6FneKdHXPxR5eCmBE2tZtXJQeg8ciS9JtAj0CVFsp
         OvRdfp9yKElNYXZTillficxRkN9ksowJDE55T2Id5kvOh1iPgyKMlb8u3m4evdZNNDa0
         2LU+ZFmy9sJs1comWtyDlMyniGKyDDhYGMoLQ=
MIME-Version: 1.0
Date: Thu, 18 Feb 2010 17:36:50 +0200
Message-ID: <468b21171002180736v543fb106w39dd7259a22d1f82@mail.gmail.com>
Subject: Performance with Multiple Node Vs Single Node
From: Richard Grossman <richiesgr@gmail.com>
To: cassandra-user@incubator.apache.org
Content-Type: multipart/alternative; boundary=001485f78aee5137cf047fe1bc86

--001485f78aee5137cf047fe1bc86
Content-Type: text/plain; charset=ISO-8859-1

Hi,

What the impact on performance to have multiple node ??

The test we run here with multiple node seems a VERY bad performance
comparing to the same amount of data located on 1 single node.
Could you confirm or infirm such statement ?

I mean if the same operation (write or read) take 5-10 times more time when
multiple node than single node where is the problem (communication between
computer) ??. I've read a lot about tuning of memory, disk, io, top, htop
but nothing about networking operation between node. Is there any way to
check what going on regarding node intercommunication ??

In my case all the queries made doesn't use 3% of the CPU nor the read
operation a lot on the disk.
Important I'm not talking about heavy load but on single query.

Thanks

--001485f78aee5137cf047fe1bc86
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">Hi,<br><br>What the impact on performance to have multiple=
 node ?? <br><br>The test we run here with multiple node seems a VERY bad p=
erformance comparing to the same amount of data located on 1 single node. <=
br>
Could you confirm or infirm such statement ? <br><br>I mean if the same ope=
ration (write or read) take 5-10 times more time when multiple node than si=
ngle node where is the problem (communication between computer) ??. I&#39;v=
e read a lot about tuning of memory, disk, io, top, htop but nothing about =
networking operation between node. Is there any way to check what going on =
regarding node intercommunication ??<br>
<br>In my case all the queries made doesn&#39;t use 3% of the CPU nor the r=
ead operation a lot on the disk. <br>Important I&#39;m not talking about he=
avy load but on single query.<br><br>Thanks <br></div>

--001485f78aee5137cf047fe1bc86--