Isn't the unpredictable sharing of objects within a single isolate a larger problem than just storag

Isn't the unpredictable sharing of objects within a single isolate a larger problem than just storage caching scenarios tho? You have the same problem without storage in the picture. Any data structures allocated by the app (ws-connection-level stuff, encryption artifacts, things coming in from input or external fetch) are going to use memory, and it won't be good if your scenario is architected to work in dev on Monday, runs fine in production on Tuesday, but then fails on Wednesday because more object instances are now put into the same isolate. Say what you want about lambda, I've used it from the very beginning, and one nice property is that it's predictable about what you get in the container you specify, and you can architect within the advertised limits.

JJohn Spurlock Isn't the unpredictable sharing of objects within a single isolate a larger prob...

kenton•7/27/21, 3:55 PM

So, an important design goal of durable objects is that it should scale to extremely large numbers of fine-grained objects. We strongly encourage developers to design for finer granularity, and we intentionally don't bill on the number of objects because we'd rather see people building apps around more, smaller objects.

If we said that each object gets its own isolate, that would unfortunately blow up this goal, since even though isolates are much cheaper than containers, they are still much more expensive than what we'd like to see for fine-grained objects.

What we really want to be able to do is adaptively adjust the isolate count to match what the app needs, so apps don't need to worry about it. We're still working on that, though, and it's a lot harder for large / coarse-grained objects compared to fine-grained.

Vvans163 i am aware of doing inside the worker ``` let { readable, writable } = n...

kenton•7/27/21, 3:56 PM

Same code should work in durable objects.

Vvans163 but how would i forward the POST request into the DO? same as a websocket? and ...

kenton•7/27/21, 3:58 PM

Your stateless worker can directly forward a request it received to a DO, like where

request

request

is the original request it received...

kenton•7/27/21, 3:59 PM

just

kenton•7/27/21, 3:59 PM

that's fine, streaming will work

kenton•7/27/21, 4:00 PM

see https://developers.cloudflare.com/workers/learning/using-streams

Vvans163 any idea then on the DO end how to get a callback for data chunk on the body? g...

kenton•7/27/21, 4:00 PM

is a

ReadableStream

ReadableStream

. https://developer.mozilla.org/en-US/docs/Web/API/ReadableStream

ReadableStream - Web APIs | MDN

The ReadableStream interface of the Streams API represents a readable stream of byte data. The Fetch API offers a concrete instance of a ReadableStream through the body property of a Response object.

Vvans163 oh wait this probably wont work because the DO will die once it exceeds the CPU ...

kenton•7/27/21, 4:25 PM

Yes, the CPU limit only get refreshed each time a new request arrives. Stream chunks don't count and new requests. Currently the limit is 500ms, but will be extended to 30s when we start billing -- maybe that's enough?

Kkenton I guess it kind of sucks that pricing ends up warping the implementation like th...

Wallacy•7/27/21, 4:42 PM

I did a similar math on my side and im keeping the memory cache for the same reason. I read 100-1000x more from memory than from storage accounting everytime the object is (for any reason) reinitialized. If read from cache can have a lower price tier i will probably remove my memory map.... apps are constrained by cost all the time, thats not a surprise.

Kkenton Yes, the CPU limit only get refreshed each time a new request arrives. Stream ch...

Wallacy•7/27/21, 4:45 PM

Thats explain few erros that i got... Nice to know!

kenton•7/27/21, 5:44 PM

@vans163 I think your code will throw an exception on the message, because

value

value

will be

undefined

undefined

, so when you try to log , it'll throw.

kenton•7/27/21, 5:45 PM

ohhhh, also, if the client just disconnects without sending a zero-size chunk to terminate the stream cleanly, then itself may throw rather than return a result.