Customer has 2 MTP with TIM installed, and want to have them an a fault tolerance configuration.
There is documentation or best practice about that?
Anyone has addressed this issue?
Thanks in advance
Since you have a CA email, you can use APM EF, IM EF, or the IM/APM Yammer groups
I'm trying to see everything that needs to be fault tolerant
- MTP only
- Tim Collector
Based on your answer, I can determine where this should go/ be answered
The component they want to be fault tolerant is theTIM (MOM already has a fault tolerante architecture ).
They have 2 MTP that will be used only for HTTP transaction inspection using TIM.
Typically this is something customers set up. So one system is on and the second is off. Some point they load and test the second one and then it is ready for failover. Companies are doing it.
I am including JMertin Guenter_Grossberger hikodavis CFAndrews to see if they have any docs on this.
Alternative is to have the Mirror-Provider architectured right for failover.
It provides data to the currently active TIM. This way - all you have to do is to configure both TIM's as regular Monitors in the CEM setup, and only the traffic that flows will determine which one is active.
IMHO the best approach. Make sure - that the same traffic is never sent to both TIM's at the same time!
Customer has an aggregator taps that send traffic to MTPs, if I understood, you say to configure taps to automatic switch the traffic to another tap's port (that is connected to the secondary MTP) only when the tap's "primary port" doesn't feel Ethernet signal up on the MTP endpoint? (primary).
Yes. That is correct. That is one way of doing it.
Have you any additional information about other ways to do it ? (automatically failover is preferred if possible otherwise manually)
Implementing something like the port failover, delegating to the aggregator to send traffic to only primary MTP and to secondary MTP only when the primary MTP has the NIC port down, has the weakness that when the primary MTP has a software problem and doesn't works but the NIC port is up, we can lose traffic.
So does any other way of doing it. But the one described by me is the one where you loose fewest traffic data - if at all.
Sure, Mirror-Provider architecture could help with Hardware failures, but when there are software failures we need a clustering solution or an ad-hoc implementation that help to understand when the TIM is not working, otherwise the network devices continue to send traffic to the wrong TIM.
I'm thinking to use TIM monitor to understand when one TIM is working or not.
Anyone has a more smart solution to propose?
Retrieving data ...