[j-nsp] MX RE how fast is slow

[Mike Williams]

On Thursday 08 September 2011 16:01:04 Jared Mauch wrote:
> > Hmm, could you during this operations run 'top' in CLIa and couple times
> > 'show chassis routing-engine' (and maybe even 'show chassis tfeb' just in
> > case) just to see if what you're doing is CPU bound.

I've done "show chassis routing-engine" many many times.
It spikes User to >90% for a short while (a minute or so), then stays >90% 
kernel until it's finished. There are occasional spikes in User load, but 
only very brief, I figure that's it dealing with a peer/snmp/jflow/etc.


mikew at h6edccrt# show | compare
[edit policy-options prefix-list public-routersv6]
+    <blah>/48;

[edit]
mikew at h6edccrt# commit
commit complete

[edit]
mikew at h6edccrt# run show chassis routing-engine
Routing Engine status:
    Temperature                 43 degrees C / 109 degrees F
    CPU temperature             54 degrees C / 129 degrees F
    DRAM                      2048 MB
    Memory utilization          76 percent
    CPU utilization:
      User                      91 percent
      Background                 1 percent
      Kernel                     7 percent
      Interrupt                  1 percent
      Idle                       0 percent
    Model                          RE-MX80
    Start time                     2011-08-07 00:00:04 UTC
    Uptime                         32 days, 15 hours, 28 minutes, 35 seconds
    Last reboot reason             Router rebooted after a normal shutdown.
    Load averages:                 1 minute   5 minute  15 minute
                                       0.27       0.19       0.20

[edit]
mikew at h6edccrt# run show chassis routing-engine
Routing Engine status:
    Temperature                 44 degrees C / 111 degrees F
    CPU temperature             55 degrees C / 131 degrees F
    DRAM                      2048 MB
    Memory utilization          78 percent
    CPU utilization:
      User                       1 percent
      Background                 8 percent
      Kernel                    91 percent
      Interrupt                  0 percent
      Idle                       0 percent
    Model                          RE-MX80
    Start time                     2011-08-07 00:00:04 UTC
    Uptime                         32 days, 15 hours, 31 minutes, 5 seconds
    Last reboot reason             Router rebooted after a normal shutdown.
    Load averages:                 1 minute   5 minute  15 minute
                                       1.33       0.64       0.38

[edit]
mikew at h6edccrt# run show chassis routing-engine
Routing Engine status:
    Temperature                 44 degrees C / 111 degrees F
    CPU temperature             54 degrees C / 129 degrees F
    DRAM                      2048 MB
    Memory utilization          77 percent
    CPU utilization:
      User                       0 percent
      Background                 1 percent
      Kernel                     0 percent
      Interrupt                  0 percent
      Idle                      98 percent
    Model                          RE-MX80
    Start time                     2011-08-07 00:00:04 UTC
    Uptime                         32 days, 15 hours, 37 minutes, 46 seconds
    Last reboot reason             Router rebooted after a normal shutdown.
    Load averages:                 1 minute   5 minute  15 minute
                                       0.31       0.73       0.54

Maybe 8 minutes.


tfeb remained all but the same throughout.

[edit]
mikew at h6edccrt# run show chassis tfeb
TFEB status:
Slot 0 information:
  State                                 Online
  Intake temperature                 41 degrees C / 105 degrees F
  Exhaust temperature                57 degrees C / 134 degrees F
  CPU utilization                    12 percent
  Interrupt utilization               2 percent
  Heap utilization                   54 percent
  Buffer utilization                 13 percent
  Total CPU DRAM                   1024 MB
  Start time:                           2011-08-07 00:02:31 UTC
  Uptime:                               32 days, 15 hours, 35 minutes, 11 
seconds


> > I'm not entirely convinced that all blame can be put on RE, maybe JNPR
> > has less optimized code for PPC than Intel, some timing issues etc.
> > RSP720 has less beefy CPU but feels lot snappier with RIB.
>
>         When you said repopulating above, is this shown as a slowly
> updating or flushing of the krt queue?

By "repopulating" I probably mean refreshing or revaluating. I don't really 
know what it's actually doing though, which isn't helped by the RE being too 
busy for RIB interrogation.
One of the policies does happen to involve sending all 727k routes through a 
series of local-preference changes based on AS path and communities, and 
refreshing RIBs (if it needs to).

All 3 RIBs do say they contain 727k BGP routes, with 364k active.
Is 2.1M RIB routes too many? :)

-- 
Mike Williams