Computing On The Edge - 202337
It Is Not About Speed
Spreading computation across a cluster can be about being faster, processing more, and in less time. Sometimes, it's not about speed, it's just a necessity, because you wouldn't be able to cope with large amounts of data otherwise.
Not necessarily this is top priority. Sometimes the real advantage of spreading data across nodes does not lie in performances, but on resilience.
The worst thing with traditional data storage is Single Point Of Failure.
Everything works, but if something happens, you are instantly offline.
Reality is not binary, a lot of things can go wrong, and experience tells us that planning for some disruptions is a good thing. Everything can fail, and eventually, it will.
Plan for it. Do not put all your eggs in one basket. Some temporary service disruption is just fine, if the entire business does not blows away with it.
Reading Tips for the week
Distributed Computing
Stream Processing vs Real-time OLAP vs Streaming Database
The Way To Go
Implementing a distributed key-value store on top of implementing Raft in Go