Recently, we've been growing like crazy and that's been putting some extra stress on our infrastructure and creating more transient errors than we'd like. Most of these errors, don't actually have any impact on customers as the process is just retried.
However, a few of these have created some very brief outages in the past couple of weeks that have effected our customers' websites for very short periods, like 10 seconds. We hate this.
After a lot of research and analysis we found a key bottleneck in how we were sending data between our cloud server applications. We realized that we needed to ease the heavy load on this crucial API data pathway. To do this we've added some additional caching to the requests going through this bottleneck.
The result of this solution, is a dramatic reduction in error rates and resource usage, which has in turn created and dramatic increase in performance, that is seen most in the load times for our IDX on our customer websites. The improvement in page load speed we've seen is profound.