Lanem wrote:What is the best way to get a specific timing data value out of a load testing experiment. I want to run up to 10,000 simultaneous users and I'm OK with using cloud machines to do the work.
There are lots of hard ways to do this, but the EASY way is the SaveRecord command.
The reason this is hard is that to get 10,000 users running eValid instances will requore about 10 "large" machines -- typically you rent these short term from the cloud, e.g. Amazon Web Services -- and about 10 separate user accounts on each of these machines. It is possible to network these, but it is simpler [and we think better as well as easier] to just let all 10,000 eValid's write the key facts into local files on each machine and then collect the data into one place for analysis.
We've used this in the past, with good success. There is a slight risk that some of the data which are supposed to be near-simultaneous may not be if some machine doesn't have the correct universal time to the nearest second...but our experience is that even with a dozen machines they all know what it is quite accurately.
The other thing we do is make sure that the first records of the SaveRecord output is a timestamp that is sortable. By that we mean that if you simply concatenate all of the files together and sort from the left hand side, then all of the records will be organized along the common timeline. This makes the data -- and with 10,000 users all running even as few as 10 repetitions that would be 100,000 records -- very easy to analyze for trends because it is all in time-sequential order.
The key to this is the SaveRecord Command
that reads key timer and byte-could information from inside an eValid browser instance, and writes it out to a named file. This output, as you can see from the command explanation, is going to be a simple text file. The key parameter to do the timestamp described above is "%T"; the output from this parameter is precisely in the form such that a left-to-right sort will put all the records in the right order.
--eValid Loading Team