Hi all,
Based on statements by Storj staff, I know the official Storj satellites run on instances across multiple Google Cloud Platform regions and AZs. Does Storj (the company) has any future plans to run their satellites across multiple independent cloud providers to avoid single points of failure?
I ask because the service Railway recently had an incident where Google mistakenly disabled their GCP account and disrupted their entire service for several hours. Although a variety of their services were also running on AWS and on their own infrastructure, they all depended on functionality provided by GCP and so that incident took down everything.
Thank you.
A few recent commit I see they migrating to TiDB, it could be a good move, as for now, google have them in their pocket, where do you use cloud if your data is in GCP (spanner)? I just hope they keep postgres/cockroachdb as currently as I’m using GitHub - oxidecomputer/cockroach: CockroachDB 22.1.x long-term maintenance branch · GitHub
Actually we uses several providers, not only GCP. And @kocoten1992 is correct, we are doing a research how to make it more independent.
The Spanner is not the only distributed DB in the world, it’s just faster and a little bit cheaper than competitors with the comparable set of required features, but of course with caveats. The support of PgSQL and CockroachDB should be continued as far as I understand.
Have you try YugabyteDB yet? Would really love to hear about quirks of database 
It’s in the list of consideration, however, as far as I can see TiDB is the leader for our conditions. We will see.
Maybe it would be published 
Interesting, and that’s very forward-looking.
Out of curiosity, are things set up such that if GCP were to do something similar to what they did with Railway (i.e., taking down one’s entire GCP account and all services under that account) that there’d be minimal-if-any disruption, or are things still dependent on GCP and it’d take some time to restore?
Is there any consideration to running Storj-owned hardware in several colo facilities somewhere rather than relying on cloud infrastructure for the satellites?
Storj traditionally doesn’t own hardware, we are renting it from different providers, like from any SNO, to do not have a single point of failure.