On Sat, Sep 05, 2009 at 10:58:20PM -0400, Derek J. Balling spake thusly:
> The raw-bandwidth capacity of SATA-II is 3Gbps, or about 300MB/s.
> Even if I concede 50% overhead on that, that's still greater than
> the capacity of a single GigE.

Right. But no SATA-II disk I have ever owned has come anywhere near
saturating that. 3Gbps is pure marketing.

> Which is why SANs (from my experience anyway) tend towards
> FibreChannel (either 2Gbps or 4Gbps) or multiple Ethernet
> connections bonded together via 802.3ad/LACP/etc. protocols (like
> you're doing).

Which is why I typically bond at least two.

> But a single ethernet channel is *definitely* a bottleneck for most
> disk access....

Most? Seems pretty unlikely to me. Depends on the workload. Even disk
to disk local RAID rebuilds don't do 100MB/s. Have you actually
measured how many megabytes per second you are really transferring
over your SAN? I bet it isn't nearly as much as you think. I have
munin graphs which show that my SAN (running various MySQL, http,
email, etc servers on Linux) rarely needs even 100MB/s. And I'm
handling hundreds of thousands of emails and tens of thousands of
unique IPs/web hits a day.

-- 
Tracy Reed
http://tracyreed.org

Attachment: pgpj0B7kGNPEt.pgp
Description: PGP signature

_______________________________________________
Tech mailing list
[email protected]
http://lopsa.org/cgi-bin/mailman/listinfo/tech
This list provided by the League of Professional System Administrators
 http://lopsa.org/

Reply via email to