Cisco gurus:
I configed slb on two 6509s (native IOS, version 12.1(12c)E5). The performance is disappointing. The traffic to load balaced servers is about 4M/s. Previously, cpu load was less than 5%, now it's 30%. sh run sometimes takes 10 minutes to show up. A lot of messages like below come up. Our hardware (SUP1A, MSFC2 and PFC)dictates that IOS 12.1(12c)E5 may be the highest version that we can run.
I can only find one technical tip regarding SLB on 6500, and it's quite basic. Seems not too many people use this feature.
I post this to ccie-lab@groupstudy.com, and one guy replied that he is running almost identical config with Sup2, and he has no problem at all.
Can someone shed some light?
Thanks,
Weidong
--------------------------------
Cat6509#sh run
...
ip slb serverfarm TESTFARM
probe PROBE-PING
! by the way, http probe doesn't work!
!
real 192.168.66.234
inservice
!
real 192.168.66.235
inservice
!
ip slb vserver TEST-VS-FTP
virtual 192.168.75.163 tcp ftp service ftp
serverfarm TESTFARM
sticky 3600
inservice standby V-SERVER
!
ip slb vserver TEST-VS-WWW
virtual 192.168.75.163 tcp www
serverfarm TESTFARM
sticky 3600
inservice standby V-SERVER
Cat6509#sh ip slb reals
real farm name weight state conns
-------------------------------------------------------------------
192.168.66.234 TESTFARM 8 OPERATIONAL 533
192.168.66.235 TESTFARM 8 OPERATIONAL 411
Cat6509#sh log
....
%ICC-SP-5-WATERMARK: 1004 pkts for class TCAM-API are waiting to be processed
%IPC-SP-5-WATERMARK: 990 messages pending in rcv for the port Card17:Request(10000.8) seat 2210000
%ICC-SP-5-WATERMARK: 949 pkts for class TCAM-API are waiting to be processed
%IPC-SP-5-WATERMARK: 931 messages pending in rcv for the port Card17:Request(10000.8) seat 2210000
%IPC-SP-5-WATERMARK: 870 messages pending in rcv for the port Card17:Request(10000.8) seat 2210000
%ICC-SP-5-WATERMARK: 865 pkts for class TCAM-API are waiting to be processed
%IPC-SP-5-WATERMARK: 806 messages pending in rcv for the port Card17:Request(10000.8) seat 2210000