In-memory storage for benchmarking

[haraldng]

Hello,

I'm master student doing my thesis where I investigate in performance of distributed consensus algorithms. In the documentation it says that the in-memory storage cannot be bootstrapped. Is there any way to run a LogCabin cluster with in-memory storage for benchmarking?

[ongardie]

Hi @haraldng, I don't know if there's an easy way to do that. It's sort of an unusual configuration, since most people want their data to be durable. Here are a couple of ideas:

1. You could easily run with Segmented storage against a tmpfs.
2. You might also be able to bootstrap the cluster with a Segmented storage member, then add a bunch of Memory storage members, then remove the original Segmented storage member.

[haraldng]

I will try using tmpfs. Thank you for the advice!

[haraldng]

Hi again,

Is the writeEx function used in the Benchmark example a blocking function? Blocking in terms of that it only returns to the client when a majority of Raft servers have committed that write operation?

[ongardie]

Is the writeEx function used in the Benchmark example a blocking function? Blocking in terms of that it only returns to the client when a majority of Raft servers have committed that write operation?

Yep. The client library is meant to be pretty hard to use unsafely.