Re: Extended RAC - latency on real systems
Date: Thu, 3 Apr 2008 11:17:44 +0200
Dnia czwartek, 3 kwietnia 2008 06:50, Matthew Zito napisał:
> You'd be better off asking a question like this on the NANOG mailing list:
> There's nothing specific about RAC versus any other traffic-based latencies
> for various packet sizes. One thing to keep in mind for your research,
> though, is that most extended-distance WAN links have smaller MTUs, and
> hence jumbo frames are not an option.
As far as I know, most Extended RAC configurations are running on dedicated DWDMs or pure Gigabit Ethernet/FC optical links (2-20km). I think that in this second scenario MTU of Ethernet can be set as high as 9k.
Thanks for link, I'll try to ask there if no one replies here.
> -----Original Message-----
> From: oracle-l-bounce_at_freelists.org on behalf of Jakub Wartak
> Sent: Wed 4/2/2008 8:40 PM
> To: oracle-l_at_freelists.org
> Subject: Extended RAC - latency on real systems
> Hello experts,
> I'm doing small academic research about Extended RAC and latencies, could
> anybody send me round-trip-tipe average latencies (with distances and
> interface types) observed between distributed nodes? Preferably measured
> using ICMP ping with standard packet sizes and bigger ones (up to MTU, e.g.
> ~9000 for Jumbo Frames, on Linux it's the "-s <size>" argument). Last two
> lines of /bin/ping would be pretty enough...
-- Jakub Wartak http://vnull.pcnet.com.pl -- http://www.freelists.org/webpage/oracle-lReceived on Thu Apr 03 2008 - 04:17:44 CDT