4 ms·
It's warm - when you change the ImageUri or "Update Code" for the lambda definition, it downloads the container into "somewhere" lambda-y - this takes a few sec
by abofh 4y ago
It's warm - when you change the ImageUri or "Update Code" for the lambda definition, it downloads the container into "somewhere" lambda-y - this takes a few seconds depending on size. Startups are fairly quick, but because of the way it persists your running image in memory, your container is generally (on frequent usage) 'paused' between invocations, and resumes quite quickly.
- Kalanos 4y agoso cost-wise that's probably equivalent to running a server? but you get all of the nice lambda integrations like APIGW
- adamgordonbell 4y agoCost wise you are just paying for the time and memory when the lambda is running. So the image may be 'close' to the lambda but it's billed like other lambdas.
- mjb 4y agoYeah, that's right. I talk about some of the details here: https://www.youtube.com/watch?v=A-7j0QlGwFk https://www.youtube.com/watch?v=A-7j0QlGwFk We're working on some more formal documentation of this, both to help customers understand how to optimize things, and to join the research conversation about container loading. Our solution combines deterministic flattening, demand loading, erasure coding, convergent encryption, and a couple of other cool technologies to both accelerate the average case and drive down tail latency.
- abofh 4y agoThanks for the share, useful to have some of the details explained!