Shouldn't all of those tasks be done before you image is baked and ready to run in prod? You can definitely write all the assets to a separate disk layer and not ship the uneeded tools.
-
-
Replying to @MylesBorins @jasnell and
sometimes? not for a bunch of the things we currently automate in lambda
2 replies 0 retweets 0 likes -
Replying to @brianleroux @MylesBorins and
Cloud functions are a product you buy. Tell them to sponsor some Open Source work in this area. We will be very happy to work with folks from Microsoft, Google, AWS to improve these metrics. They never brought this up or flagged this to be a problem.
2 replies 0 retweets 0 likes -
Replying to @matteocollina @brianleroux and
I did a talk at NINA last year on some of the issues with Node cold start in general on serverless. Most of them are mitigated if you can build and webpack/bundle your code into a single file. It's pretty easy to get seconds of cold start due to just file IO, even locally.
2 replies 0 retweets 2 likes -
Replying to @crandycodes @matteocollina and
I haven't done any analysis of cold start over the versions of Node.js, though. It might be interesting, but it hasn't come up as an issue. Biggest issue is just module load times.
@brianleroux, you're mostly talking about npm start up time, though, right?1 reply 0 retweets 1 like -
Replying to @crandycodes @matteocollina and
no, the Node container. There's benchmarks in the thread. That said, npm has bloated significantly and thats an issue which will impact container perf for sure. It got mentioned because I think it is the quickest first win.
2 replies 0 retweets 1 like -
Replying to @brianleroux @crandycodes and
I think this awesome feedback for
@npmjs. Truly, this is not@nodejs or the TSC fault. Happy to help.1 reply 0 retweets 2 likes -
Replying to @matteocollina @crandycodes and
Well, sure, that's one win. The benchmarks of regressions of coldstart is a Node problem. The narrative that it has been improving is false and should be concerning. It is to me!
1 reply 0 retweets 0 likes -
Replying to @brianleroux @matteocollina and
@MylesBorins the raw perf numbers from acmeair are trending up (see latency and throughput). But start/stop time has trended up. Node.js started pretty small so its not a large absolute change but it is an upward trend. See: https://benchmarking.nodejs.org/charts/start_stop.png ….1 reply 0 retweets 0 likes -
Replying to @mhdawson1 @brianleroux and
We don't have a graph for footprint on disk, but I'd be happy if somebody had time to contribute a test/scripts to add that.
1 reply 0 retweets 0 likes
I'd be interested. Is there an example or existing PR for that kind of thing you can point me at?
-
-
Replying to @southpolesteve @brianleroux and
It would be less complicated that this one, but its probably the best example: https://github.com/nodejs/benchmarking/pull/123 … The key is that a script to run the bench is added, and then the chart framework is configured to display the data.
0 replies 0 retweets 2 likesThanks. Twitter will use this to make your timeline better. UndoUndo
-
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.