Really depends on the use case and the traffic pattern. I probably have a strong personal bias for DynamoDB, but if it's high scale and read-heavy (which is the usual case), then it really is probably the best option since it also handles provisioning / partitioning for you. If you need a stronger consistency level for some reason, postgreSQL/mySQL+vitess works. If it's write heavy for some reason, then Cassandra would be a good choice. If it needs to support text search, that'll require an inverted index, which makes ElasticSearch/OpenSearch a good choice, but a few other options like Redis and postgreSQL also support using inverted indexes.
It used to be pretty consistent at 10:30pm PST / 11:00am IST for around 6 months straight, but I'm currently experimenting bit with other times that are more friendly for all timezones. I'm currently trying to post a placeholder for upcoming streams at least 3 days in advance, like you might see for some VTuber channels, but again, just experimenting with some options right now
i went throught your video, tbh its not begineer friendly, you could have done a little go through about what commentId, threadId and parentId means, how many levels of replies are we expecting? where the bottleneck lies, i know you are a very high level staff engineer in some FANG all your colleagues think you are a genius, and all respect for that, but tbh these videos are so non intuitive, i wish you put more hardwork in it, your views might also increase drastically, i think you can even afford an IPAD and make illustrations using pencil,else dont make these low effort videos, they are just wasting youtube storage
There's plenty of beginner-friendly content out there for system design problem coverage already, nothing for people looking to grow to senior+ though. Literally several whole books by people like Alex Xu and Stanley Chiang, but not a single applied problems book currently on the market that mentions causal consistency, sequential consistency, or sharded counters. In addition, I have the _only_ coverage on the entire internet for at least 6-12 problems out of the 70 that I've covered. Probably another one or two dozen for which I have the only comprehensive coverage, given that all the interview books currently out there only cover a net of roughly 30-40 distinct problems You're correct that it's not beginner-friendly and there's a lot of room for improving quality, but that's a very intentional trade-off that I made for trying to efficiently spend my time. I am planning to take a multi-pass approach and come through with iterative improvements to quality and depth, with a preference for depth -- but in the meantime, I am extending my warmest invitation for you to go cry a river 💌 Really though, if you're lost while trying to watch my content, just go read Alex Xu's books. They're great.
Starts on 14:39
what do you mean by DAG comments?
Nice idea using parent id for a comment to keep track of nested comments. What DB would you recommend for this use case?
Really depends on the use case and the traffic pattern.
I probably have a strong personal bias for DynamoDB, but if it's high scale and read-heavy (which is the usual case), then it really is probably the best option since it also handles provisioning / partitioning for you.
If you need a stronger consistency level for some reason, postgreSQL/mySQL+vitess works.
If it's write heavy for some reason, then Cassandra would be a good choice.
If it needs to support text search, that'll require an inverted index, which makes ElasticSearch/OpenSearch a good choice, but a few other options like Redis and postgreSQL also support using inverted indexes.
Thank you for taking out time to explain. Keep creating content like this.@@SDFC
very useful
please keep making such interesting and useful content
When do you host these sessions?
It used to be pretty consistent at 10:30pm PST / 11:00am IST for around 6 months straight, but I'm currently experimenting bit with other times that are more friendly for all timezones.
I'm currently trying to post a placeholder for upcoming streams at least 3 days in advance, like you might see for some VTuber channels, but again, just experimenting with some options right now
i went throught your video, tbh its not begineer friendly, you could have done a little go through about what commentId, threadId and parentId means, how many levels of replies are we expecting? where the bottleneck lies, i know you are a very high level staff engineer in some FANG all your colleagues think you are a genius, and all respect for that, but tbh these videos are so non intuitive, i wish you put more hardwork in it, your views might also increase drastically, i think you can even afford an IPAD and make illustrations using pencil,else dont make these low effort videos, they are just wasting youtube storage
There's plenty of beginner-friendly content out there for system design problem coverage already, nothing for people looking to grow to senior+ though.
Literally several whole books by people like Alex Xu and Stanley Chiang, but not a single applied problems book currently on the market that mentions causal consistency, sequential consistency, or sharded counters.
In addition, I have the _only_ coverage on the entire internet for at least 6-12 problems out of the 70 that I've covered. Probably another one or two dozen for which I have the only comprehensive coverage, given that all the interview books currently out there only cover a net of roughly 30-40 distinct problems
You're correct that it's not beginner-friendly and there's a lot of room for improving quality, but that's a very intentional trade-off that I made for trying to efficiently spend my time.
I am planning to take a multi-pass approach and come through with iterative improvements to quality and depth, with a preference for depth -- but in the meantime, I am extending my warmest invitation for you to go cry a river 💌
Really though, if you're lost while trying to watch my content, just go read Alex Xu's books. They're great.