This is an embryonic idea that I'd like to capture here for future discussions when the post-Denali release of SSIS is in the planning stages. It was prompted by a comment to my blog post:
Is the SSIS Catalog going to be enough?
That comment was:
"The problem with the new built in logging is the fact that it is not centralized. If you have 10 servers, you will have to visit each server to look at your ssis logs."
Putting aside the question of why on earth anyone would have 10 servers it does raise an interesting point.
Currently if we wanted all SSIS logging to occur centrally (i.e. on a single server) then all projects would need to be deployed to and executed on that SSIS server. In an organisation where they have many packages running all day long it is quite possible that that SSIS server could become a CPU/Memory/Network bottleneck - its not unfeasible that that organisation would want to scale out by having multiple SSIS servers however as soon as you do that you lose the benefit of centralised logging.
I propose a solution to this below.