Mailing-List: contact user-help@hadoop.apache.org; run by ezmlm
Precedence: bulk
Reply-To: user@hadoop.apache.org
Received-SPF: neutral (athena.apache.org: local policy)
MIME-Version: 1.0
In-Reply-To: 
 <CAE_FxDhz6=1wGcgW8MPRQHnqac93WSTNLGCxaY7QiNgU25mAaA@mail.gmail.com>
References: 
 <22324_1359651197_0MHI005OM1K8FF00_99DD75DC8938B743BBBC2CA54F7224A706D67714@NYSGMBXB06.a.wcmc-ad.net>
 <CAE_FxDhz6=1wGcgW8MPRQHnqac93WSTNLGCxaY7QiNgU25mAaA@mail.gmail.com>
From: Ted Dunning <tdunning@maprtech.com>
Date: Thu, 31 Jan 2013 11:28:52 -0800
Message-ID: 
 <CAND0qzuyXoYhDcmgJ9cRYsViqkVv9qMs03pS8XAdiixpqnP2NA@mail.gmail.com>
Subject: Re: Dell Hardware
To: user@hadoop.apache.org
Content-Type: multipart/alternative; boundary=e89a8ff2534a39045c04d49aa437

--e89a8ff2534a39045c04d49aa437
Content-Type: text/plain; charset=ISO-8859-1

We have tested both machines in our labs at MapR and both work well.  Both
run pretty hot so you need to keep a good eye on that.

The R720 will have higher wattage per unit of storage due to the smaller
number of drives per chassis.  That may be a good match for ordinary Hadoop
due to the lower I/O efficiency, but with higher performance distributions
like MapR, you will probably do better on combined acquisition, ops and
power costs per terabyte with the C2100 series.

As you do your selection, you need to determine what your primary
constraint will be.  Will it be storage size?  I/O rates?  Compute load?

Once you have that figured out, the hardware selection should fall out
directly from a TCO analysis that looks at the different scenarios.

On Thu, Jan 31, 2013 at 11:13 AM, Andy Isaacson <adi@cloudera.com> wrote:

> On Thu, Jan 31, 2013 at 8:53 AM, Artem Ervits <are9004@nyp.org> wrote:
> > Does anyone run Hadoop on Dell R720 model of servers? Dell site lists
> C2100
> > model of servers as best fit for Hadoop workloads. What does community
> > recommend?
>
> The R720 supports up to 2 xeon CPUs and 8 drives in 2U. If configured
> appropriately it will work fine as a Hadoop node. For example, 2x
> Xeon, 64GB RAM, 8x 2TB SATA configured as JBOD, would be a reasonable
> worker node. Buy 20 of those and now you're cooking with gas. :)
>
> The C2100 is probably somewhat cheaper for a given storage amount
> since it fits 12 disks in a 2U chassis (so you don't need to buy as
> many machines to achieve a given amount of capacity). Dell doesn't
> provide a configurator for that model so I can't compare prices.
> Apparently it's using the previous generation of Xeon -- 5500/5600
> versus e5 on the R720 -- which probably means poorer throughput per
> watt.
>
> The right answer will depend on your application and workload.
>
> -andy
>

--e89a8ff2534a39045c04d49aa437
Content-Type: text/html; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable

We have tested both machines in our labs at MapR and both work well. =A0Bot=
h run pretty hot so you need to keep a good eye on that.<div><br></div><div=
>The R720 will have higher wattage per unit of storage due to the smaller n=
umber of drives per chassis. =A0That may be a good match for ordinary Hadoo=
p due to the lower I/O efficiency, but with higher performance distribution=
s like MapR, you will probably do better on combined acquisition, ops and p=
ower costs per terabyte with the C2100 series.</div>

<div><br></div><div>As you do your selection, you need to determine what yo=
ur primary constraint will be. =A0Will it be storage size? =A0I/O rates? =
=A0Compute load?</div><div><br></div><div>Once you have that figured out, t=
he hardware selection should fall out directly from a TCO analysis that loo=
ks at the different scenarios.<br>

<br><div class=3D"gmail_quote">On Thu, Jan 31, 2013 at 11:13 AM, Andy Isaac=
son <span dir=3D"ltr">&lt;<a href=3D"mailto:adi@cloudera.com" target=3D"_bl=
ank">adi@cloudera.com</a>&gt;</span> wrote:<br><blockquote class=3D"gmail_q=
uote" style=3D"margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1e=
x">

On Thu, Jan 31, 2013 at 8:53 AM, Artem Ervits &lt;<a href=3D"mailto:are9004=
@nyp.org">are9004@nyp.org</a>&gt; wrote:<br>
&gt; Does anyone run Hadoop on Dell R720 model of servers? Dell site lists =
C2100<br>
&gt; model of servers as best fit for Hadoop workloads. What does community=
<br>
&gt; recommend?<br>
<br>
The R720 supports up to 2 xeon CPUs and 8 drives in 2U. If configured<br>
appropriately it will work fine as a Hadoop node. For example, 2x<br>
Xeon, 64GB RAM, 8x 2TB SATA configured as JBOD, would be a reasonable<br>
worker node. Buy 20 of those and now you&#39;re cooking with gas. :)<br>
<br>
The C2100 is probably somewhat cheaper for a given storage amount<br>
since it fits 12 disks in a 2U chassis (so you don&#39;t need to buy as<br>
many machines to achieve a given amount of capacity). Dell doesn&#39;t<br>
provide a configurator for that model so I can&#39;t compare prices.<br>
Apparently it&#39;s using the previous generation of Xeon -- 5500/5600<br>
versus e5 on the R720 -- which probably means poorer throughput per<br>
watt.<br>
<br>
The right answer will depend on your application and workload.<br>
<span class=3D"HOEnZb"><font color=3D"#888888"><br>
-andy<br>
</font></span></blockquote></div><br></div>

--e89a8ff2534a39045c04d49aa437--