I have been running a azure redis cache for about 6 months now without issue. Last week I started getting timeouts on one of the client connections, so I went into the Azure portal to see what was going on. Right away I saw that the redis cache adviser was telling me that I was nearing my network bandwidth limit for of 5Mbits/sec on my C0 Standard pricing plan. So, I went and looked at the charts and noticed that about every hour there is a huge spike in cache reads that goes from my normal 188 kB/s to 1.2 MB/s??? I have checked all my clients and I do not have any jobs that run every hour that would cause this? Is it possible that the Azure infrastructure is doing this? Any ideas?
1
votes
Have you tried performing any benchmarks for your individual clients?
– Dave Cooper
We experienced a very similar (if not same) issue last week as well with a C0 on Azure. Ultimately we bumped it up to a C1, but we continued seeing strange behavior until a few days later on Friday, Nov. 18 when our Redis instance began acting more normally again. Did you ever resolve this issue on your end? We still don't know what happened.
– Rk220
I ended up upgrading as well. Once I did, the strangeness went away. $$ == fixed
– Scott Smith
1 Answers
0
votes
I ran into a similar issue right away, but in my case, it was not every hour, but every 5 min. After a while, I notice that it was because, in the client (browser) a soft-reload happened at a fix interval of 5 min for a bunch of minor global configurations (such as telemetry gathering and other info). Since my developer tests usually don't last that long, that never came-up before... so check for that too...
other than that, unless the "keep-alive" is malfunctioning the only explanation I can provide for that fact is that azure is actually grouping your data in that time-frame, so, if you had a second-by-second report, you would perhaps see that you have a wave, or a massive spike...
