Include latency and maximum concurrent connections benchmarks about frameworkbenchmarks HOT 2 CLOSED

We agree, additional metrics such as latency would be great additions. We have received a couple other comments of a similar nature via e-mail. We're not certain if we can get that kind of data out of the test tool we're using (WeigHTTP), but we'll investigate. One or two other testing tools have been recommended, so we'll look into those as well.

It won't be something we can do super quickly since Pat put effort into parsing the WeigHTTP results into JSON files that the blog entry renders. That work will need to be repeated if we change to another tool. But ultimately, if another tool gives us more and better data, we'd prefer to use that.

We do in fact capture total run time for each test and that is in the JSON data files if you look at them. We just don't render that anywhere in the blog entry. We had toyed with rendering the total "EC2 cost" for running each test on the two Large instances as an amusing factoid but passed on it in the end.

Raw WeigHTTP capture for Go (as an example):

https://github.com/TechEmpower/FrameworkBenchmarks/blob/master/results/i7/20130321085035/json/go/raw

from frameworkbenchmarks.

We're putting together the next round of numbers now, and this time we're using Wrk (https://github.com/wg/wrk) instead of WeigHTTP (https://github.com/lighttpd/weighttp), so we'll have latency numbers. We're hoping to have updated results later this week.

from frameworkbenchmarks.