3 ms·
Tim Bray: Serverless Latency?
- yarapavan 8y agoBut my personal favorite choice for serverless compute is the Go programming language. It’s got great, clean, fast, tooling, it produces static binaries, it’s got superb concurrency primitives that make it easy to avoid the kind of race conditions that plague anyone who goes near java.lang.Thread, and finally, it is exceedingly readable, a criterion that weighs more heavily with me as each year passes. Plus the Go Lambda runtime is freaking excellent.
- yarapavan 8y agoHow to talk about it · To start with, don’t just say “I need 120ms.” Try something more like “This has to be in Python, the data’s in Cassandra, and I need the P50 down under a fifth of a second, except I can tolerate 5-second latency if it doesn’t happen more than once an hour.” And in most mainstream applications, you should be able to get there with serverless. If you plan for it.
- PaulHoule 8y ago(1) For some applications it is all about the "cold start" latency. I am building up a "smart home" system and that involves the system doing quite a bit a work ahead of time so that you don't wait 0.4 seconds for a list of albums in your music system to display the first page, or for weather radar to load, etc. (2) There is a way to eliminate 20 sec waits at P100 and that is to say that you are going to abort at 1 second, not start requests which are obviously going to take >1 second, etc. This is a "path not taken" because most people would rather deal with the long waits rather than the fails, but serving one of those 20 sec requests can sometimes mean 100 other requests are late.