We are a huge production setup where it’s absolutely critical successfully but we use temporal cloud. Hosting it yourself is what makes it miserable. https://temporal.io/resources/on-demand/netflix
Cockroach doesn't offer strict serializability. It has serializability with some limits depending on clock drift. Also CockroachDB does not provide linearizability over the entire database.
You’re describing compare and swap which is a good solution. You’re pushing complexity down to the database, and remember this is distributed locking. When you have a single database it’s simple until the database crashes leaving you in state of not knowing which of your CAS writes took effect. In major systems that demand high availability and multi datacenter backups this becomings pretty complicated with scenarios that break this as well around node failure. Usually some form of paxos transaction log is used. Never assume there is an easy solution in distributed systems… it just always sucks
That’s kinda what aws tells people when its services go down. If your backend can’t take a short outage without weeks of recovery then it’s just a matter of time.
Seems like unnecessary shade to throw on the helpline. There was a viral thing a bit ago making rounds on social media accusing them of it but its far from reality. There are like 50k of 3 million calls a year that result in contacting the authorities, but its extreme scenarios.
Its a rare thing and highlighting it as the default scenario is spreading FUD and possibly keeping people from getting the help they need. It is true though that if you call and describe an immediate plan to kill yourself there is some legal, ethical, and liability requirements that mean they will call authorities. Exact same thing if you tell your doctor or therapist.
Xbox, steam, google play store, Nintendo and sony are comparable. Same percentages and everything, 30% is kinda game industry standard. Game engines starting to take bigger cut too.
It’s not a real requirement. Just a performance thing. Each partition has an index and as it grows the deserialization of it to find the queried start column takes longer. It is lazy loads but there a perf hit. Easier to have a rule of thumb like “100mb partitions” and have more fixed tps per core etc expectations.