I wanna say this kind of generosity is what makes the world a better place. Thank you so much!
@zoelkh Жыл бұрын
yeahh
@LibraryOfTheOligarchs Жыл бұрын
agreed!
@nawafb6280 Жыл бұрын
When the service XYZ is down, introducing a circuit breaker is a good idea. It helps to halt the routing of new requests to the failed service until it has recovered. Consequently, we avoid displaying the online payment option for new orders during this downtime.
@simonhuang3193 Жыл бұрын
Do the location updates also need a User-Machine Mapping data store like the one you have in the Facebook Messenger design?
@abdulhalim391510 ай бұрын
Why do we need strongly consistent reads ?
@mohammedaltaf4316 Жыл бұрын
Hlooo, make some machine learning system design videos
@SDFC Жыл бұрын
I have quite a few out now. I think I'm currently most happy with my coverage of item-item collaborative filtering in the "amazon similar item recommendation" video
@maximizedchen6875 Жыл бұрын
So, is Postgres going to be able to scale for us? Lol
@nobytes2 Жыл бұрын
I don't understand why the video assumes pg can't scale, I think that's bad information.
@SDFC Жыл бұрын
it can't scale without "manual partitioning", which is widely regarded as a fool's errand due to the heavy operational burden that it introduces when possible (via relaxed consistency and isolation requirements for the given use case), it's best to use a database solution with managed partitioning like DynamoDB (or arguably Cassandra to some extent) however, there are cases where strict isolation and consistency requirements can't be avoided (such as most variations of flash sale or ticketmaster), and that's where you're stuck with manually partitioned SQL databases There are solutions out there like Vitess that'll kinda do the partitioning for you on MySQL, but that comes with its own suite of extra trade-offs and it's best to just use a mature and fully managed solution like the ones that were mentioned above