Catalyst 4948E FIB TCAM utilization

[atis]

Hello,

  is it possible to check the FIB TCAM utilization of a Catalyst 4948E?

  There is a command on Catalyst 6500 platform (show platform capacity forwarding), but I couldn't find any similar one for Cat4948E...

BR, Attila

[Ivan Shirshin]

Hi,

That command shows the capacity and utilization of the EARLs in the system and 4500 does not use those, thus no such command. 4500 may have TCAM exhaustion in QoS or ACL TCAM though:

http://www.cisco.com/en/US/products/hw/switches/ps663/products_tech_note09186a008054a499.shtml

show platform hardware acl statistics utilization brief

Kind Regards,
Ivan Shirshin

**Please grade this post if you find it useful.

[atis]

Thanks, I knew this url, but I am looking for fib tcam utilization.

AFAIK the Cat4948E supports 57,000 routes, but I don't know how it is partitioned (ipv4, ipv6 and multicast) and the current usage will be interesting as well...

-A.

[Vignesh Rajendran Praveen]

Hi Atis,

The FIB TCAM utilisation on a Cat4948E can be seen using the below command.

show platform hardware ip route summary

****** Plz do rate this post if rou found it useful ***********

Thanks & Regards,

Vignesh R P

[atis]

Thanks, this is what I need.

I started to decode the output of this command. Could you help me to translate some things?

show platform hardware ip route summary

block#  start   end     mode    entries used    free    group   type

0       80 Bit  0       4095    4096    4096    0       3       Dst

1       160 Bit 4096    8190    2048    419     1629    6       Dst

2       80 Bit  8192    12287   4096    0       4096    0       Src

3       80 Bit  12288   16383   4096    0       4096    0       Src

4       80 Bit  16384   20479   4096    4068    28      3       Dst

5       80 Bit  20480   24575   4096    0       4096    0       Src

6       80 Bit  24576   28671   4096    3692    404     3       Dst

7       Unused  28672   32767   4096    0       4096    -       -

8       Unused  32768   36863   4096    0       4096    -       -

9       Unused  36864   40959   4096    0       4096    -       -

10      Unused  40960   45055   4096    0       4096    -       -

11      Unused  45056   49151   4096    0       4096    -       -

12      Unused  49152   53247   4096    0       4096    -       -

13      Unused  53248   57343   4096    0       4096    -       -

14      Unused  57344   61439   4096    0       4096    -       -

15      Unused  61440   65535   4096    0       4096    -       -

group#  inUse   mode    type      lookup  entries free    util%   rangeId

0       yes     80 Bit  uRPF Ipv4 Src     12288   12288   0       0

1       yes     160 Bit uRPF Ipv6 Src     0       0       100     1

2       yes     160 Bit SpecSrc   Src     0       0       100     255

3       yes     80 Bit  UC Ipv4   Dst     12288   432     96      0

4       yes     160 Bit SpecDst   Dst     0       0       100     255

5       yes     160 Bit MC Ipv4   Dst     0       0       100     1

6       yes     160 Bit UC Ipv6   Dst     2048    1629    20      2

7       yes     320 Bit MC Ipv6   Dst     0       0       100     3

...

As I see group#3 is for unicast ipv4, the utilization of this group is 96%, which is very high. It means if the switch receives more ipv4 routes, then these routes will be software switched (or dropped)? Or is it possible to change the size of these groups?

-A.

[bissaha]

HI 

Once its full, you might get following errors

"%C4K_L3HWFORWARDING-2-FWDCAMFULL: L3 routing table is full.  Switching
to software forwarding" 

Yes it will then be software switched and might end up getting High CPU.

The cat4k is limited to around 128k routes in TCAM resources only.

Please check the section "UNICAST Forwarding:CEF"   under following slide from networkers

http://www.cisco.com/networkers/nw04/presos/docs/RST-3508.pdf

Hope this helps

BISWA

[Vignesh Rajendran Praveen]

Hi Atis,

As Biswa says when more ipv4 routes are recieved and reaches 100% utilisation, traffic would be software switched resulting in High CPU issues. And moreover it is not possible to change the size of these groups.

****** Plz do rate this post if you found it useful ***********

Thanks & Regards,

Vignesh R P

[atis]

Hello,

  despite of the group#3 is utilized ~96%, there are more unused blocks (from group7 to group15), So I think these blocks can be allocated to group#3, which allows 4k routes per block in case of ipv4 routes. In case of unicast ipv6 routes there is only 2k routes/block, and assigned to group#6. Am I right?

BR, Attila