Discussion

Chas and Ryan are continuing work on the survey. Scott seems to be tied up, but Ken is applying some pressure about shib as use case. may look for some participation from salsa too. We feel good about the survey as it stands so far, but we need a broader audience. We're seeking out more people to ask for input on the spreadsheet.

Russ reported that they are starting to get a lot of data out of projects like PIPES. Lots of network measurement data. Want to get it into a common format as defined by ggf [mark: I don't remember this reference]. one problem we have is that we're generating massive amounts of data but not much of it is actually interesting. What really constitutes an event? they want to kick out an event when a measurement is outside a normal envelope, and figuring out where that envelope is will be variable on the infrastructure and the environment.

What if a user is having problems with an app, and they want to know how the network was doing at that time? they'd look at event records. This is the point of making the network data available to the middleware diagnostic backplane.

Mark wondered whether this one might be a test case that would be easier to put together. To the extent that e2e performance measures might use shib to control access, we might kill two birds with one stone - use it as a test case for diagnosing when Shib might break down for this application - diagnosing the difficulty. Problems in the measurement/collection system (holes in data) may mask or obscure problems in the network.

Russ will think about writing up a use case as consumer and will send status on where he is with use case as producer and what the next steps are.

General energy and interest for Middleware diagnostics are good, but we need bodies (volunteers to help). We need to turn interest into more people.