APA - OUTPUT LINK-MONITOR

[Mario_88]

Hello:

Is the load balancing algorithm "perfect" or it should show some differences between the ports? this is the output of link-monitor in my system
port#./link-monitor -n 10 -t 60 -p 900

Display stats accumulated at intervals of 60 seconds for
inbound/outbound frame count, bits/sec, and errors/discards.

Gathering data for first interval.
Please wait 60 seconds for the first output....
IN OUT IN OUT IN OUT
PPA packet packet b/sec b/sec Error/Dscrd Error/Dscrd
May 23 18:18:02 900 110.2K 119.3K 1.7M 6.7M 0 2 0 0
> 11 72.1K 73.8K 1.1M 3.7M 0 0 0 0
> 13 38.1K 45.6K 586.6K 3.0M 0 0 0 0
May 23 18:19:02 900 217.9K 236.4K 1.7M 6.8M 0 4 0 0
> 11 149.1K 151.8K 1.2M 3.8M 0 0 0 0
> 13 68.8K 84.6K 537.8K 2.9M 0 0 0 0
May 23 18:20:02 900 363.8K 394.8K 2.1M 7.3M 0 7 0 0
> 11 257.5K 259.1K 1.5M 4.0M 0 0 0 0
> 13 106.3K 135.7K 588.8K 3.3M 0 0 0 0
May 23 18:21:02 900 521.2K 570.3K 2.3M 8.5M 0 8 0 0
> 11 352.5K 353.8K 1.6M 4.4M 0 0 0 0
> 13 168.6K 216.5K 753.8K 4.1M 0 0 0 0
May 23 18:22:02 900 636.7K 699.1K 2.3M 8.5M 0 9 0 0
> 11 427.1K 430.4K 1.6M 4.4M 0 0 0 0
> 13 209.6K 268.7K 749.8K 4.1M 0 0 0 0
May 23 18:23:02 900 724.7K 797.5K 2.2M 8.3M 0 10 0 0
> 11 486.4K 493.2K 1.5M 4.5M 0 0 0 0
> 13 238.2K 304.3K 714.3K 3.9M 0 0 0 0
May 23 18:24:02 900 820.1K 897.8K 2.1M 7.9M 0 10 0 0
> 11 538.4K 553.1K 1.4M 4.4M 0 0 0 0
> 13 281.7K 344.6K 714.3K 3.5M 0 0 0 0
IN OUT IN OUT IN OUT
PPA packet packet b/sec b/sec Error/Dscrd Error/Dscrd
May 23 18:25:02 900 960.9K 1.0M 2.2M 7.9M 0 11 0 0
> 11 630.9K 654.9K 1.5M 4.6M 0 0 0 0
> 13 329.9K 389.0K 728.8K 3.3M 0 0 0 0
May 23 18:26:02 900 1.1M 1.1M 2.2M 7.5M 0 24 0 0
> 11 698.3K 725.7K 1.5M 4.5M 0 0 0 0
> 13 357.3K 418.8K 708.4K 3.1M 0 0 0 0
May 23 18:27:02 900 1.2M 1.3M 2.2M 7.4M 0 37 0 0
> 11 784.3K 816.7K 1.5M 4.5M 0 0 0 0
> 13 384.1K 444.9K 684.1K 2.9M 0 0 0 0


MAX MAX
IN OUT
PPA b/sec b/sec
May 23 18:27:02 900 2.3M 8.5M
> 11 1.6M 4.6M
> 13 753.8K 4.1M

[Steven E. Protter]

Shalom,

There is no such thing as a perfect algorithm.

Your data though not a large sample proves that.

If you sample for a longer period of time under heavy load you will find the data is distributed equally assuming the two NIC cards are identical and there are no problems with configuration on the switch.

Switch configuration plays a big part in how well APA works.

SEP

[Mario_88]

Hi,

I have attached the output of link-monitor during two hours.

Regards.

[Mario_88]

Using load balancing mode we saw 4-5% percent of packte loss after switching to hot standby the packet loss was less than 1%