from Brent Hedden – Program Analyst III FileMaker Certified Developer (8,11-13) Clinical Trials Unit (CTU), University of Louisville, Kentucky, USA
Aug 6th, 2015
By email
"Good day Nick!
I'm writing to you in response to a video that Richard Carlton released recently about FM Server Performance Improvements.
.
It was mentioned that you created a way to simulate multiple FM clients to be able to load test a server. Can you or would you be willing to share with me how exactly you did this? This would be very handy to have, and FileMaker Inc. has been no help to me in this regard, and I'm not finding any other ways through the FM community.
"I can easily vouch for this! I've been looking for a way to accurately simulate multiple FileMaker clients for a few years now. With the Perform Script On Server feature (which is what this solution is mostly using), this makes accurately testing server performance a breeze.
I'm currently testing migrating from a physical Mac Mini, FMS 13 to a virtual server, Windows OS, FMS14. Using this tool has given me the ability to not only compare the abilities of the two servers, but also to adjust the virtual enviroment for optimal use for my usage.
It also has the ability to instantiate the concept of tokens, if you wanted to see if this will work well in your environment."
October 25th, 2015
from Brent Hedden by email
"I’ve been mainly interested in using this application as a means of upgrading a Mac Mini running FMS 13 to a virtual server running MSWindows Server 2012 with FMS 14. I’ve experienced major issues with transferring FMS to a virtual environment before, so this time I wanted to make absolutely sure that this will be a smooth transaction.
Although I didn’t fully test things as I wanted, the dsBenchmark tool gave me some great metrics on comparing v13 to v14.
I especially found the max load feature great for stress testing.
One item I still have to test is how changing the amount of CPU cores and available RAM affects the overall performance. With having a virtual environment, these hardware changes can be made very easily.
My current plan with using the dsBenchmark tool is to run all of the tests over again. Since it was a few months since I’ve done this, I noticed some inconsistencies with the way that I ran the testing procedures. Now that I’ve run the utility several time, I know what to expect of it and how to use it to my advantage.
Thanks for keeping in touch with me. I don’t have a current timeframe, but my retesting is at the top of my wordlist now. Once I’ve compiled the results, I’ll be more than happy to share the results with you."
October 28th, 2015
I've finally been able to find enough free time to run the tests that I wanted. Attached is all of the charts, exported data, and an Excel sheet with the data from FMS logs. Here is a breakdown of what I tested:
FMS 13 on a Mac Mini, 10.9.5, 2.6 Ghz i7 CPU, 16GB memory, SSD drive. Cloned machine to an identical hardware, and installed FMS v14 for comparison. I couldn¹t run tests from Webdirect, as the cloned machine didn't have a signed certificate(different IP address), and v14 pretty much requires this now. Most users are FMP clients, so this wasn't critical.
Ran 2 different tests on each machine - a mix of users for 15 minutes, and as many intense users until peak.