New Metric Visualizations

[CannonLock]

Collection of charts and the issues they would have helped debug.

---

People have been reporting issues with slow transfers and I as a system administrator want to see what namespace is putting the most pressure on the system.

Stacked histogram where the area of each bar chart corresponds to number of transfers and bytes transferred by namespace. X axis is time.

[CannonLock]

@bbockelm Just requested a time series of the IO for caches so that we can see what load is crippling our caches. We should create a timeseries graph that is grouped by server so we can see what caches specifically are being hammered and when.

[brianhlin]

As an origin admin, I would want to be able to identify if the bottleneck is in the filesystem or the origin. So maybe something like IO read/write bytes/s and network in/out bytes/s. Also, error rates for each layer would be helpful (scitokens vs multiuser vs ofs [?])

[CannonLock]

What about the interaction of IO and network would help you determine if it is the filesystem or the origin?