search

mikhailshilkov / mikhailio-hugo

Sources of https://mikhail.io migrated to Hugo
MIT License
12 stars 8 forks source link

Comments to "Eliminate Cold Starts by Predicting Invocations of Serverless Functions" #40

Open mikhailshilkov opened 4 years ago

mikhailshilkov commented 4 years ago

Add your comment to Eliminate Cold Starts by Predicting Invocations of Serverless Functions.

The comments will be displayed directly on the page. I may edit or merge your comments in a non-significant way to make sure they are displayed properly on the website.

nothingto commented 4 years ago

Thanks for posting this. I’m looking forward to seeing your tests to these changes - i’ve been following all of your functions benchmark posts (across providers such as AWS, Azure, and GCP). If I can request, can you please test these changes with C# also in addition to javascript? Looking forward to the results!

I use HTTP functions for use cases such as being a backend to mobile or web. This means invocations are often very consistent one after another. I’d love if you test a scenario where functions are invoked frequently (like for example, during the business hours, functions receive frequent invocations, but then fall off during off-hours and become infrequent).

SimonEismann commented 4 years ago

Very interesting blog post!

I feel like the optimization will do comparatively little to mitigate coldstarts. As you mentioned, the only functions that benefit are rarely executed functions that do not really use more than one instance. To me its hard to find an example for a function with that workload pattern that highly cares about latency.

However, I think its still a great optimization due to the following point: Because for many applications the keep-alive interval can drastically reduce... The proposed optimization likely highly reduces the time instances are kept-alive without ever receiving requests, which seems beneficial for Azure.