Tuesday update on slowness issues. Still making small incremental gains. Current hypothesis: Network routing may play a role.

 Over the weekend, Tyler Technologies shut down their system briefly to do some performance tuning on the EFM API. Now that we've got a full day of logs to examine since the most recent change, we can report that there has been a 16% decrease in the average response time compared to last Monday.😊

That's good news, and worth celebrating, but that's still over 3 times the average response time prior to their cloud migration, so this is still a top priority for us.

We've been a bit frustrated by the fact that it was a change to someone else's system that caused this. That means we're in the awkward position of just measuring slow response times and complaining instead of being able to contribute to a solution. However, one of our operations folks pointed out that network latency is something we can assess from our side of the connection, so we've been doing that and are finding some definite problems that we can use to start some diagnosis from our side of the connection. We're working that angle now, and have passed along our findings to Tyler in the hope that it will help.