Brian Armstrong
@barmstrong
Auto-scaling web servers is common. But I don't think I've seen anyone auto-scale databases before? (someone must have done it, but we couldn't find much) Anyway - I gave our Platform team the challenge and they built it! Awesome work. https://t.co/iBly7ECtB7
7 replies
36 recasts
106 reactions
Mats β DuneCon πΉπ
@mats
Dune autoscales our DuneSQL query engine (without any ML, just by looking at incoming requests)
0 reply
0 recast
3 reactions
emilee
@emilee
really cool stuff! curious -- if aws aurora let you easily plug in the predictive indicators y'all developed to trigger autoscaling, would that have also solved for this use-case?
0 reply
0 recast
0 reaction
Chinmay πΉοΈπΏ
@chinmay.eth
CC @amccausl @kijijij
0 reply
0 recast
0 reaction
K
@kijijij
That's great, I used double derivative of load to predict increase / decrease in traffic, worked much better than deducing "factors" associated with load.
0 reply
0 recast
0 reaction
3obby
@3obby
A double linked list on an EVM kinda auto-scales :)
0 reply
0 recast
0 reaction
Ivy
@ivy
neon's been doing it with serverless postgres
0 reply
0 recast
0 reaction
atlasmoth
@osuji.eth
@barmstrong does the team use k8s for auto scaling the DB or sth else? Also, the blog says the new model relies on volatility from external price data, is there any sort of smoothening that's performed on the data first? And finally, will the code be open source eventually?
0 reply
0 recast
0 reaction