super sick, would love to hear some

super sick, would love to hear some details

6 Replies

AJR•9mo ago

Lots of work done on more efficiently hitting cache, more aggressively releasing connections to origin back into the pool, and tuning timeouts across our systems so that you're less likely to see a cold start under common usage patterns. Just general improvements across the board.

moody 💭OP•9mo ago

that's cool although not sure how that results in the improvement to client write performance

AJR•9mo ago

Releasing connection more aggressively means more are available for use, resulting in overall increases in throughput/latency on concurrent requests. Fewer cold starts means more requests have a prewarmed connection and will be much faster. Just off the top of my head. Or am I misunderstanding what you're observing in terms of difference?

moody 💭OP•9mo ago

i see, what's the max throughput you think? would you be able to max out the 10 gbps internet connection of an rds instance?

AJR•9mo ago

That's a good question. It'd depend on a lot of factors, but probably not today if I had to guess. We can definitely max out the NIC on a workstation and residential Internet. For more complete benchmarking of data/storage products, stay tuned!

Hebilicious•9mo ago

Performance question, is this useful to do release hyperdrive like this on each request?

const db = getDb()
const response = await getResponse()
context.waitUntil(db.end());
return response

const db = getDb()
const response = await getResponse()
context.waitUntil(db.end());
return response

I guess I should post this in the general channel

Gaming

Programming

super sick, would love to hear some