@Motroskin said in #66:
> Fair point. So, you think this is just a budget concern?
I can't think of any reason other than keeping the budget tight or trying to avoid complexity, but I'm not part of the admin/devel team.
Thanks for all your efforts and keep up the good work.
Don't listen to some whiners that want to get 99.9999% uptime for a FREE service run mostly by volunteers. Get a life people. One full day downtime in total, over 14 years existence. Don't change anything if it costs too much. I can perfectly live with these rare downtimes and will always come back.
Thanks for the clear explanation
Thanks a whole lot for everything you do - lichess is a testament to what the internet can be at its best and I absolutely love it. The people who keep it running are doing the world a great service that goes beyond chess.
So there's this cool technology called virtualization. Check it out. You can host your machines in the cloud as VMs. No dependency on physical hardware. That move will go a long way toward making Lichess more redundant in the future.
@rushkeldon said in #13:
> I would have thunk that Lichess was running on AWS or Google Cloud.
> Is it perhaps the cost that makes that path less interesting / viable?
> I would think that most of the "required physical action in the datacenter" types of issues would not be present in a cloud environment.
”Cloud environment” is just someone else’s datacenter / computers.
@robertl30 said in #76:
> So there's this cool technology called virtualization. Check it out. You can host your machines in the cloud as VMs. No dependency on physical hardware. That move will go a long way toward making Lichess more redundant in the future.
pondering or brainstorming that.
Which cloud now? Is this not just displacing the risk. Maybe that cloud service could be a contract promise to keep that VM going when their hardware gets broke, they would have the redundancies for it. This is just about software independence, the hardware problem might be the same. Is this not just paying someone to do things behind our management. It might have its equivalent costs. Just curious, I don't know that scale of things but I think virtual does not change the hardware layer issues.
It might make changing provide of the hardware more easy? As Dockers might do. No clue curious. to learn, the mechanisms at play.
@gpa150chess said in #29:
> Agree, people who recommend these providers only have a superficial understanding about how things are run.
> Another thing I was wondering: I'm not sure if this played a role as well for thibault and his team: OVH is a european/french company. Considering lichess' general attitude and values I believe it definitely played a factor in choosing OVH over the the likes of google, amazon etc.
I have very good understanding of cloud providers. Using some random cloud provider in the name of some random principles is disrespectful to the user. This outage is a perfect example of that, the bigger cloud platforms have very high uptimes and quicker response to outages
Even longer than Fire in a Lichess data center