DX - Load Balancing & Application Acceleration
DX - Load Balancing & Application Acceleration

show capacity vs. show vmstat

‎10-21-2008 11:49 AM

We have 3 DX-3600 appliances running in active/active mode and we might need a 4th incase one appliance goes down. I am trying to find out the amount of free resources during our peak usage, but when I run show capacity and show vmstat I am getting two different results. The cpu usage on show capacity is the amount of cpu being used and the cpu usage on show vmstat "n" is the amount idle, but even taking this into consideration they are still showing different results.

 

Which one shows the correct information?

 

^Clb-jun-dx-01% show cap

  CPU      Memory                         Network
                             ether 0                      ether 1      
(% used)  (% used)    (bytes in)    (bytes out)    (bytes in)    (bytes out)
   46        44           0             0             0             0      


lb-jun-dx-01% show vmstat 5
 procs      memory      page                   disk   faults      cpu
 r b w     avm    fre  flt  re  pi  po  fr  sr ad0   in   sy  cs us sy id
 1 1 0 1141940 1925816    5   0   0   1   0   0   0  166   77   3 32 77 -9
 2 2 0 1086084 1925812  194   0   0   0   0   0   0 7919 96978 4622 18 49 33
 2 2 0 1086084 1925812  182   0   0   0   0   0   0 7995 66721 5207 12 28 60
 2 1 0 1141940 1925812  206   0   0   0   0   0   0 7827 83585 4602 15 42 44 

1 REPLY 1
DX - Load Balancing & Application Acceleration

Re: show capacity vs. show vmstat

‎10-22-2008 07:08 AM

The 'show capacity' command shows values based off an average from the vmstat output:

 

CPU - show % in use by userland and system (average over the last 60 seconds.).
Memory - Amount of free memory in the system (average over the last 60 seconds).
Network - show bytes in per second for interface ether0 and ether1 (average over the last 60 seconds).

 

 

So these displayed values are an average over the last minute and the sample interval is one second. This allows smoothing of  the peaks in the values, as previously using just vmstat output intermittent CPU spikes would cause the DX to report the CPU usage as "red" for a brief period.  There is a CLI command "set server factory a ssw" to change the (default) average period to less than  60 seconds.

 

 

 

Announcements

DX SERIES

The Juniper Networks DX application acceleration platform delivers a complete data center acceleration solution for Web-enabled and IP-based business applications.

RSS Icon