[c-nsp] can't be -- output discards on 6500 gig-e

[Steve Francis]

If your 1 minute average is 500Mbps, then your peaks are certainly
higher. I suspect TAC is correct.


Stephen Sprunk once said on this list:

One should note that any utilization up to 59.8% is, on average,
indistinguishable from an empty line.  70% = 1.6x delay, 80% = 3.2x, 90%
= 8.1x, and 95% = 18.05x.  Of course, once you figure in finite
buffering, anything past 59.8% is likely to be dropping packets.

ObMath: Plot r^2/(1-r).  Where the derivative exceeds one (r~0.598),
delay increases faster than traffic rate.  Assumes random arrival times.






> -----Original Message-----
> From: cisco-nsp-bounces at puck.nether.net 
> [mailto:cisco-nsp-bounces at puck.nether.net] On Behalf Of Edward Henigin
> Sent: Thursday, July 29, 2004 9:37 AM
> To: cisco-nsp at puck.nether.net
> Subject: [c-nsp] can't be -- output discards on 6500 gig-e
> 
> Hello,
> 
> I've got a 6509/sup720 that's having output discards on a 
> gigabit ethernet interface at peak time.  No biggie, you say, 
> you're running the line too hot.  But what's "too hot?"  
> We're seeing the discards when the interface is doing 500Mbps 
> on a 1 minute average.  I do not believe at all that we 
> should be getting output discards at that traffic level.
> 
> The card is a WS-X6416-GBIC.  It has 10 ports enabled.  We 
> constantly saw discards on the port at a rate of 40 to 70 
> packets/sec over a 3 hour period.  During the same 3 hour 
> period, the port was doing 450Mbps to 650Mbps, 40kpps - 
> 55kpps, on a 5 minute average.
> 
> We opened a case with TAC and all they say is "you're running 
> out of output buffers."  I'm sorry, but that's not helpful at 
> all.  This behavior is unacceptable to me.  At 500Mbps, I 
> should NOT be discarding packets, even at that low rate.
> 
> Send flowcontrol is on, receive flowcontrol is off.  By my 
> reading of http://tinyurl.com/6omgc, that means that we will 
> not process incoming pause packets, and we should not be 
> pausing our output to the other side.
> So I don't believe that flowcontrol has anything to do with it.
> 
> Anyone else have any insight?
> 
> Thanks,
> 
> Ed
> _______________________________________________
> cisco-nsp mailing list  cisco-nsp at puck.nether.net 
> https://puck.nether.net/mailman/listinfo/cisco-nsp
> archive at http://puck.nether.net/pipermail/cisco-nsp/
>