[c-nsp] carving TCAM on Engine 3 linecards due to exhausted IPv6 TCAM regions
Jan Sandmaier
sandmaier at schlund.net
Fri Feb 18 06:49:50 EST 2011
Hi Per,
ok, after reading this again, the counters don't show a mis-function.
Thanks,
Jan
Am 18.02.2011 11:20, schrieb Per Carlson:
> Hi Jan.
>
>> But if I look into TCAM usage I see no difference/improvement except as
>> for the RX_IPv6_128 region:
>>
>> #execute-on slot 1 sh controllers frfab alpha tcam carve
>> ========= Line Card (Slot 1) =========
>> Id Region % Curr/Carve/Dflt Mask Blocks Masks
>> Value Cells
>> Total/Used(%) Total/Used(%)
>> Total/Used(%)
>> -----------------------------------------------------------------------------
>> 7 RX_IPv6_128 10.13/14.00/ 4.00 830/4 0.48% 1660/8 0.48%
>> 26560/128 0.48%
>> 8 RX_IPv6_127 0.01/ 0.00/ 0.00 1/0 0.00% 2/1 50.00%
>> 32/16 50.00%
>> 9 RX_IPv6_126 0.03/ 0.00/ 0.00 3/3 100.00% 6/6 100.00%
>> 96/96 100.00%
>> 10 RX_IPv6_125 0.01/ 0.00/ 0.00 1/0 0.00% 2/1 50.00%
>> 32/16 50.00%
>> 11 RX_IPv6_124 0.01/ 0.00/ 0.00 1/0 0.00% 2/1 50.00%
>> 32/16 50.00%
>> 14 RX_IPv6_121 0.01/ 0.00/ 0.00 1/0 0.00% 2/1 50.00%
>> 32/16 50.00%
>> 28 RX_IPv6_107 0.01/ 0.00/ 0.00 1/0 0.00% 2/1 50.00%
>> 32/16 50.00%
>> 29 RX_IPv6_106 0.01/ 0.00/ 0.00 1/0 0.00% 2/1 50.00%
>> 32/16 50.00%
>> 55 RX_IPv6_80 0.01/ 0.00/ 0.00 1/0 0.00% 2/1 50.00%
>> 32/16 50.00%
>> 71 RX_IPv6_64 0.04/ 0.00/ 0.00 4/3 75.00% 8/7 87.50%
>> 128/112 87.50%
>> 79 RX_IPv6_56 0.01/ 0.00/ 0.00 1/1 100.00% 2/2 100.00%
>> 32/32 100.00%
>> 86 RX_IPv6_49 0.01/ 0.00/ 0.00 1/0 0.00% 2/1 50.00%
>> 32/16 50.00%
>> 87 RX_IPv6_48 0.96/ 0.00/ 0.00 79/77 97.46% 158/154 97.46%
>> 2528/2464 97.46%
>> 88 RX_IPv6_47 0.02/ 0.00/ 0.00 2/1 50.00% 4/3 75.00%
>> 64/48 75.00%
>> 89 RX_IPv6_46 0.02/ 0.00/ 0.00 2/2 100.00% 4/4 100.00%
>> 64/64 100.00%
>> 90 RX_IPv6_45 0.01/ 0.00/ 0.00 1/1 100.00% 2/2 100.00%
>> 32/32 100.00%
>> 91 RX_IPv6_44 0.03/ 0.00/ 0.00 3/3 100.00% 6/6 100.00%
>> 96/96 100.00%
>
> The 128 region holds all TCAM resources for all IPv6 prefix lengths.
> Resources are automatically moved to other prefix lengths when needed.
>
> From the Bug-description (referenced before as well):
>
> "For the purposes of IPv6 scalability, the RX_IPv6 regions are the
> ones we're interested in. Initially, all IPV6 cells are placed into
> the /128 region. As we add other mask lengths into the ipv6 routing
> table, we will "move" blocks of the TCAM into a new region that
> corresponds to the mask length for that prefix."
>
>
More information about the cisco-nsp
mailing list