Hi, is it possible to build a private network of Storj nodes (2 or 3) and access them trough S3 API for a personal use? Or must we use Tardigrade for that purpose?
If yes, is there any guide to build it and give it a try?
Thanks.
You can run your own satellite, but you would probably need more nodes.
Do you intend to offer random people to run the nodes or will those nodes be entirely under your control?
If entirely under your control, would’t something like ceph or glusterfs be better as it does not involve paying the nodes etc (in which case you would be paying yourself) and is probably faster.
Thanks. As i’m discovering the product, i thought it could be run like MinIO and benefit from the encryption+erasure parts.
Main goal was to run a completely private and distributed system in the family, nothing more :slight_smile
A big part of Storj is the idea that the nodes are not trusted - node operators would cheat if it was possible to get more money that way. Also, the nodes get paid.
If you have some servers that you own want to use, then a regular cluster would be faster and easier to set up because you do not need to bother with paying nodes (yourself), getting money from your customers (again, yourself). You also know that you will not cheat yourself 
Running four or so nodes, is probably too small a number for Storj to work properly anyway, but it would be enough for something like glusterfs or ceph (which also have erasure codes). I do not know about encryption, but you can always encrypt the data stored on the cluster.
4 Likes
It wouldn’t really work with 2 or 3 nodes, because you need 96 nodes (I think it’s 96 or might be 48, I cannot remember now…) to split the pieces of a file to provide redundancy and decentralization.
Sounds like you don’t want the redundancy or decentralization but just the encryption.
Instead of trying to create a private version of the idea, why not try StorJ as a customer?
3 Likes
I am a customer and running into a lot of problems.
Can you explain which exactly?
Cache not working (support ticket) and problems similar to those described in Github using the Gateway (gateway eating 100% CPU, errors…).
I’m actually evaluating MinIO S3 gateway using encryption; not distributed, but working “out of the box”. 
Hello,
Please go through this: https://storj.io/blog/2019/12/hosting-multiple-storage-nodes-on-the-storj-network/
I not sure it will be possible to build a private network but this link will give you a good information.
Thanks & Regards,
Anita
If your number of gateway points are low, using something like syncthing for the glue between the nodes is a perfectly acceptable, mesh, way to do things. As well, this can be directed to only operate without DHT and only over VPN tunnels to deal with the in-flight tunnel encryption. As for each location’s redundancy, that’s dealers choice… ZFS is always a good option if you set it up right and you grow in predictable and repeatable chunks, but things like Ceph work more on the scalable side if you know you’ll be growing frequently and at varied rates. Either way- you grow each node the same as each has a copy.
All of that “works” out of the box as well.