[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[TIGGE #SOD-321591]: Brazil feed

Doug,

The proposed modifications change the TCP buffer values system-wide.
The larger values will help buffer higher latency connections and will 
have no impact for connections not needing that much buffering.

mike

> Will this modification interfere with any of our other input feeds?
> 
> On Jun 25, 2007, at 2:57 PM, Unidata IDD TIGGE Support wrote:
> 
> > Doug,
> >
> > Due to the high latency, you should double these values currently
> > set on ultrazone;
> >
> >> ndd -get  /dev/tcp tcp_recv_hiwat
> > 524288
> >> ndd -get /dev/tcp tcp_xmit_hiwat
> > 524288
> >>  ndd -get /dev/tcp tcp_max_buf
> > 8388608
> >
> > and restart the LDM.  I would expect some improvement, but we may
> > need to know
> > more about the capacity of various network links along the way.
> >
> > Waldenio -- I recall you've done quite a bit of tuning on your end
> > as well?
> >
> > BTW - we became aware of a problem in the path through CLARA last
> > Friday as
> > seen by large numbers of connection failures from UFRJ to our IDD
> > cluster.
> > Tom tested the path from various points along CLARA and found
> > issues transferring
> > small amounts of data, as seen by ftp.
> >
> > mike
> >
> >> Hello,
> >>
> >> We're getting high latencies between Brazil and our server.  Have you
> >> explored this in
> >> the past, and do you have any suggestions on how we might improve
> >> network
> >> performance?
> >>
> >> http://www.unidata.ucar.edu/cgi-bin/rtstats/iddstats_nc?EXP
> >> +ultrazone.ucar.edu+LOG
> >>
> >> Thanks,
> >> Doug
> >
> >
> > Ticket Details
> > ===================
> > Ticket ID: SOD-321591
> > Department: Support IDD TIGGE
> > Priority: Normal
> > Status: Closed
> >
> 
> 


Ticket Details
===================
Ticket ID: SOD-321591
Department: Support IDD TIGGE
Priority: Normal
Status: Closed