M-Pixel
commented
3 years ago Very interesting, thanks for posting this.
I found some info here: https://blocksandfiles.com/2020/07/24/ovhcloud-acquires-openio-to-build-best-object-storage-service/
Denel said in prepared remarks: “All of the technologies used to build our Object Storage solution are and will continue to be open-sourced, which is an essential common value for both OVHcloud and OpenIO. More than that, it will be extended and enriched by our combined talents.”
...
“The OpenIO brand will disappear, but the project continues. And it will grow, taking advantage of OVHcloud’s know-how to industrialise it.”
So, if they succeed in being true to their word, we will see the name of the product change, but we will also see it continue to exist, and continue to be open-source. There is clearly a lull, as you have observed, but there is the potential that after this lull they will be able to improve at an even faster pace.
They are not paying much attention to their blog, but this is understandable. In their position, I wouldn't want to spend any resources on PR until the merge/rebranding plan is entirely solid, and I would expect that process to take at least 1 full year if they do it right.
Also, I found out that two of OVH's datacenters were massively damaged by fires earlier this month. They have been slowly but surely restoring hardware since then. I wouldn't be surprised if this slows down all R&D, including OpenIO.
On the other hand, the OpenIO Slack is still active, answering users' questions for free, and there are still regular updates to this codebase.
As a current user, I hope that they take an extremely careful approach to reconciling OIO with whatever OVHcloud has. Ideally, any optimizations or feature additions are QA'd against massive and diverse redundant/verifiable workloads for a span of months before they are promoted to release branches. Inasmuch, I would be happy if this repo's release branches saw nothing but occasional bug-fixes for the next year.
I've been running a cluster since January (about 3 months at the time of writing) with daily use, and it has yet to fail me. In fact, during that time, I've screwed up my filesystems at least twice, and OpenIO restored them perfectly.
Among the few resilient + distributed JBODs, OpenIO is very unique as a hybrid OSS/Commercial offering. There are a "bajillion" commercial-only alternatives, most of which require you to email them to negotiate a custom price that will end up being tens of thousands of dollars. CEPH and SeaweedFS are comparable, and fall in the pure-OSS camp (MinIO doesn't qualify because it requires drives to all be the same size). CEPH is notoriously over-complex and slow, has over 200 times more active issues than oio-sds, and has a backlog of almost 600 open pull-requests. SeaweedFS is maintained almost entirely by a single developer, and I ran into corruption problems when I tried it out earlier this year. As such, it would be a terrible loss if OpenIO is no longer maintained, so I hope that somebody from the company weighs in on this issue.
I ask myself a few questions, and I'm currently refraining from using this technology because I worry, it's not going to be supported in the near future. Are my worries incorrect.
If you look at the blog, it's behind the aquisition, If you look at the people who worked there before, several of them are not longer there. So my question is this going to be deprecated. If not, I have a lot more questions. But that is my main one given the look of the website, and a little bit of digging.