This week I’d like to talk a little bit about an inGraphs consolidation that is used relatively infrequently: Histogram.
Backing up a step: inGraphs provides a handful of different ways to consolidate data in meaningful/helpful ways. Consolidation can either be defined in the dashboard or manipulated using the inGraphs UI; see:
Most of these consolidation options are pretty straightforward. Aggregate (sum each series), AggregateAll (sum the sums), Max (the maxima of each series), etc. *[Side note: If you can tell me what the utility/use case for AverageAll is I’d love to hear it.] * Histogram consolidation is a little more esoteric, but sometimes it can be super-useful so I think it’s worth taking a look at…so let’s dive in!
First off: it’s not going to provide you with a histogram in the traditional sense. If you’re looking for a bar graph of a probability distribution…well…I’m sorry, folks, that isn’t a “thing” in inGraphs. What it will give you is a nifty percentage breakdown for a given metric. One of the more accessible examples I can think of is stickyrouting site traffic.
Here’s the inGraph for main partition requests per fabric:
…and here’s what it looks like if you switch consolidation to “Histogram”:
The first graph: per-fabric QPS. The second: percentage of overall per-fabric QPS.
Pretty nifty, eh? …and without having to do any “fancy” RPN magic or anything.