[Gluster-users] Switch recommendations

Dan Bretherton d.a.bretherton at reading.ac.uk
Mon Jan 30 11:58:13 UTC 2012


Hello Brian,
Thanks for that advice.

> If you already have 5524s then they are apparently stackable using a HDMI
> cable, have you tried that?
Actually we have 5424s at the moment, and a LAG is the only option for 
improving performance with those I think.
> I have bad experience with Powerconnect 5524's, but maybe you're just not
> tickling the same bugs I found (e.g.  multicast problems, failing to learn
> MAC addresses)
Thanks for that warning.  The Netgear is more expensive but probably 
worth it by the sounds of it.

-Dan.

On 01/27/2012 01:59 PM, Brian Candler wrote:
> On Fri, Jan 27, 2012 at 01:04:39PM +0000, Dan Bretherton wrote:
>> I need to buy a bigger GigE switch for my GlusterFS cluster and I am
>> trying to decide whether or not a much more expensive one would be
>> justified.  I have limited experience with networking so I don't
>> know if it would be appropriate to spend £500, £1500 or £3500 for a
>> 48-port switch.  Those rough costs are based on a comparison of 3
>> Dell Powerconnect switches: the 5548 (bigger version of what we have
>> now), the 6248 and the 7048.
> If you already have 5524s then they are apparently stackable using a HDMI
> cable, have you tried that?
>
> I have bad experience with Powerconnect 5524's, but maybe you're just not
> tickling the same bugs I found (e.g.  multicast problems, failing to learn
> MAC addresses)
>
> Haven't had these same problems with Netgear GSM7224V2, but then maybe they
> have a different set of bugs :-)
>
>> Some of the users'
>> applications do lots of small reads and writes, and they do run much
>> more slowly if all the servers are not connected to the same switch,
>> as is the case now while I don't have a big enough switch.
> If it's consistently a little slower for each individual file read and
> written, then it sounds like this usage pattern is extremely
> latency-sensitive.  If it seems to be in bursts then maybe you are getting
> small amounts of packet loss, which is causing TCP to back off.
>
> Side note: if you are latency-sensitive then don't even think about using
> 10Gig over RJ45 - it actually has a higher latency than 1G.  But 10Gig over
> SFP+ is fine.
>
> Regards,
>
> Brian.



More information about the Gluster-users mailing list