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 paolo.crosato@targaubiest.com
 designates 213.199.154.83 as permitted sender)
From: Paolo Crosato <paolo.crosato@targaubiest.com>
To: "user@cassandra.apache.org" <user@cassandra.apache.org>
Subject: RE: hardware sizing for cassandra
Thread-Topic: hardware sizing for cassandra
Thread-Index: AQHPzFhc3vopB+nQmUqGX9C0Zr6wRpv5Lx4AgAADq4M=
Date: Tue, 9 Sep 2014 19:52:42 +0000
Message-ID: <1410292361656.15345@targaubiest.com>
References: 
 <CAMFcGX3EDx1t6FyuNfpho28CRoLav8ocgdzKZAJY2phfEgk19A@mail.gmail.com>,<5BDBBC4B-8A9D-4928-BAAD-BCF7D21C4565@blackbirdit.com>
In-Reply-To: <5BDBBC4B-8A9D-4928-BAAD-BCF7D21C4565@blackbirdit.com>
Accept-Language: it-IT, en-US
Content-Language: en-US
Content-Type: text/plain; charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
MIME-Version: 1.0

Every node should have at least 4 cores, with a maximum of 8. Memory should=
n't be higher than 32g, 16gb is good for a start. Every node should be a ph=
isical machine, not a virtual one, or at least a virtual machine with an ss=
d hd subsystem. The disk subsystem should be directly connected to the mach=
ine, no sans or fiber channel between. Cassandra is cpu and io bounded, so =
you should get the maximum io speed and a reasonable number of cores.=0A=
=0A=
Number of nodes should be 3 at least with replication factor of 2. You shou=
ld prefer more less powerful nodes then fewer more powerful nodes.=0A=
=0A=
Disk size depends on your workload, although you should always keep 50% of =
the disk free in the case repair sessions requires space, or perform sub ra=
nge repairs.=0A=
=0A=
In my experience a 1GB link between nodes is ok, but the less lag the bette=
r.=0A=
=0A=
Summing up if you need to save some money, get 4 cores and 16 gb or ram, 32=
 is rarely needed and 64 a waste. 8 cores would probably be too much with 1=
000 writes a second.=0A=
=0A=
Paolo=0A=
=0A=
=0A=
...........................................................................=
.............=0A=
Paolo Crosato=0A=
Software engineer/Custom Solutions=0A=
=0A=
=0A=
________________________________________=0A=
Da: Chris Lohfink <clohfink@blackbirdit.com>=0A=
Inviato: marted=EC 9 settembre 2014 21.26=0A=
A: user@cassandra.apache.org=0A=
Oggetto: Re: hardware sizing for cassandra=0A=
=0A=
It depends.  Ultimately your load is low enough a single node can probably =
handle it so you kinda want a "minimum" cluster.  Different people have dif=
ferent thoughts on what this means - I would recommend 5-6 nodes with a 3 r=
eplication factor.  (say m1.xlarge, or c3.2xlarge striped ephemerals, I lik=
e i2's but kinda overkill here).  Nodes with less then 16gb of ram wont las=
t long so should really start around there.=0A=
=0A=
Chris=0A=
=0A=
On Sep 9, 2014, at 11:02 AM, Oleg Ruchovets <oruchovets@gmail.com> wrote:=
=0A=
=0A=
> Hi ,=0A=
>    Where can I find the document with best practices about sizing for cas=
sandra deployment?=0A=
>    We have 1000 writes / reads per second. record size 1k.=0A=
>=0A=
> Questions:=0A=
>    1) how many machines do we need?=0A=
>    2) how many ram ,disc size / type?=0A=
>    3) What should be network?=0A=
>=0A=
> I understand that hardware is very depends on data distribution and acces=
s pattern and other criteria, but I still want to believe that there is a b=
est practice :-)=0A=
>=0A=
> Thanks=0A=
> Oleg.=0A=
=0A=