Are there any good performance benchmarks for Long...
# longhorn-storage
f
Are there any good performance benchmarks for Longhorn as VM store vs other distributed file systems such as GlusterFS and Ceph? I’m thinking about making a move to Harvester from oVirt but I’m worried about I/O performance of Longhorn especially with write intensive loads. It looks like V2 data engine is potentially around the corner but I’m not sure about timelines on that. How much more performant is V2 suppose to be?
l
Hi @future-yak-96960, here you can find a presentation with a comparison between v1 and v2 performance. And this is the roadmap .
b
There are general benchmark results in the github wiki for each minor version although 1.8 appears missing https://github.com/longhorn/longhorn/wiki/Performance-Benchmark
FWIW, we moved from ovirt to harvester. It's not running on like for like tin so a direct comparison is hard but we haven't had any disk IO issues. We use replica 1 for workloads which distribute the storage themselves (DBs etc). Although there is an issue with using replica 1 storage and 0 downtime harvester upgrades. https://github.com/harvester/harvester/issues/6145
l
There are general benchmark results in the github wiki for each minor version although 1.8 appears missing
What about the performance, the results for 1.8 should be very similar to the ones for 1.7
b
Should be but it would be nice if there were some benchmarks for it
data engine v2 keeps getting more and more functionality so that might change