Forum Discussion
NimbostratusMy LTM VE is running 2 TMM's but why? How do I determine what each TMM is for?
We have a 200 Mbps license for our LTM VE appliance. Occasionally we are exceeding our bandwidth limit and after some research I found the total bandwidth available is divided by the number of TMM's you are running. I've run the CLI command to determine I have 2 TMM's running so I am now license for 100 Mbps per TMM:
[root@F5-2:Active:In Sync] config tmsh show sys tmm-info | grep Sys::TMM
Sys::TMM: 0.0
Sys::TMM: 0.1
How can I determine what each TMM is for?
From the GUI under System > Resource Provisioning only Management (MGMT) and LocalTraffic (LTM) are provisioned resources. Are those my two TMM's? I do have IP Intelligence running on the LTM's but that was enable through the tmsh. I did not install ASM.
Thank you!
Each TMM (Traffic Management Microkernel) instance uses one vCPU so the BIG-IP can do more work by running two vCPU's in parallel. By default half of the connections are "disaggregated" to each TMM. This is a statistical multiplexing process. When there are many connections, system utilization will be balanced. If you have just one very busy connection then it may dominate the activity of the TMM handling it.
6 Replies
M_QuevedoEach TMM (Traffic Management Microkernel) instance uses one vCPU so the BIG-IP can do more work by running two vCPU's in parallel. By default half of the connections are "disaggregated" to each TMM. This is a statistical multiplexing process. When there are many connections, system utilization will be balanced. If you have just one very busy connection then it may dominate the activity of the TMM handling it.
AICadmin_2415Okay so my LTM VE does has 2 vCPU's so as you said one TMM for each vCPU assigned to it.
Obviously, I don't think the appliance would run very well dropping my vCPU count just to increase my bandwidth throughput on a single TMM.
Thank you for the input! I wish I could have found a F5 kb article that explained that.
Each TMM (Traffic Management Microkernel) instance uses one vCPU so the BIG-IP can do more work by running two vCPU's in parallel. By default half of the connections are "disaggregated" to each TMM. This is a statistical multiplexing process. When there are many connections, system utilization will be balanced. If you have just one very busy connection then it may dominate the activity of the TMM handling it.
- AICadmin_2415
Okay so my LTM VE does has 2 vCPU's so as you said one TMM for each vCPU assigned to it.
Obviously, I don't think the appliance would run very well dropping my vCPU count just to increase my bandwidth throughput on a single TMM.
Thank you for the input! I wish I could have found a F5 kb article that explained that.
Kishore_Kumar_2Cirrus
First of all, the LTM Or any other resources mentioned in the Resource provisioning tab are the modules or functional types that the BigIP can work as.
TMM is the traffic management (microkernel?), it is the brains of the BigIP. It takes up the traffic and manipulates it according to our configuration.
The multiple TMMs is because the box has multiple cluster processing that helps traffic processing
https://support.f5.com/kb/en-us/solutions/public/14000/300/sol14358.html
- boneyard
MVP
depending on your license you might be able to add cores and thus tmms.
https://support.f5.com/kb/en-us/solutions/public/14000/800/sol14810.html
@ M Quevedo does a VE also safe half of the available TMMS for mgmt handling?
