All in all it's a pretty great tool and it gives a high level of visibility into what's happening on the server side. You can look at percentiles, filter to your heart's desire, set SLA's, monitor very specific sections of traffic, etc. The UI can be a bit clunky, but once you get used to it there are a lot of options. They also have a "render time" measurement which is essentially supposed to capture the time between the start of the request and the onload event (so essentially full page load time).
Our main complaints with Coradiant are as follows:
- It can take a really long time to drill down into requests. When you stop looking at high level aggregated statistics and want to look at a group of individual requests (say all image requests taking over X seconds) the report takes multiple minutes to run, which isn't bad if you do this once in a while but definitely gets frustrating if you want to look at a lot of reports quickly. The newest version of the product attempts to remedy this with a new tool that allows you to slice data more quickly.
- Coradiant is great at letting you know when things slow down, but not great at letting you know why. This is not really a problem with the tool, it has no idea why a given packet (or set of packets) took longer to arrive, but this is the benefit that a product like New Relic provides. When you are in the middle of an incident Coradiant gives you very limited information about what is causing the problem (depending on the nature of the incident). It can report on client IP, resource size, user agent, etc. but once you start drilling into that kind of data things slow down quite a bit which makes it tough to diagnose an ongoing issue.
- Coradiant is fairly expensive, and as we move to multiple data centers and our traffic keeps increasing it is only going to get more pricey.
As companies get significantly larger and have hundreds of engineers, it can make sense to build something in house to monitor performance, but until that point Coradiant can be a good choice. For client side performance I really like tools like LogNormal and Torbit Insight, which give you more granular information about what's going on in the browser. The server side is really where we want a good, scalable tool. Like I said above Coradiant has been really great for us so far, we are just getting to the size where it makes sense to see what other people are doing to see if we can build something in house or if there are other viable options out there.