Hello all, long time lurker, first time poster. I’ve been running a single Storj node since April 2019 and have dedicated 20TB space on my server behind a 750/750 fibre Internet connection. Storj is running on Docker on Ubuntu 18.04 Server. The 20TB is on a ZFS raid array used for other purposes, and I can scale up the size of the node by managing the backend ZFS pool.
I’ve read a number of posts on this forum where different SNOs have reasons for setting up multiple nodes on the same host, and I’m now wondering whether there is any benefit to setting up additional nodes on my server given my configuration, i.e. 2x 10TB nodes, or 4x 5TB nodes.
Since in a single location, all your nodes would be behind the same IP, they will receive the same or less amount of data as they would if you were running only one single node. This is because of IP filtering - we want to create a network that is as decentralized as possible.
So I’m asking the community for thoughts and advice to get some pros and cons for switching up my setup from a single node on my server to multiple nodes. Should I keep things as is or add more Storj docker containers to my setup? Thanks in advance!
As you quoted above there is no benefit to run multiple nodes, if you have a RAID-like setup already.
With multiple nodes you just increase the consuming of resources without any benefit.
The multisetup is considered as an alternative to complicated and expensive RAID setups.
But since you already used it for other purposes, then there is no other reason to run multiple nodes.
There are no benefits from an economy standpoint to increase the node count based on your setup. This is more a factor for users who is only using a single physical hard drive per node. If that HD crashes the reputation of that node will be affected, but not the rest of the nodes. It could be seen as a way of running raid/zfs without having raid/zfs…
In your case a single node that you manage in a way that it will NEVER break will be more beneficial, but with the risk of you loosing 20TB of data.
I’ve taken the same path, running my 20TB node with multiple Raid-5 with StoragePools to allow for expansion or “adjustments”. Battery on the Raid Cards and UPS for the server and storage array. Feel pretty confident about the setup. The only change I might want to do is to virtualize the CPU part, that would allow for hardware redundancy towards the storage array
the only difference is when you runing RAID you always loose in space. if you have 20TB space in raid then it will be about 25-30 when it no raid. All the difference. But today no one know how it will be filled.
Also no raid is faster when it needed fast Egress.
Completely understood about the loss of space on my RAID setup due to parity protection. That’s a drag that I’ve chosen to be okay with since I host other data on this array for non-Storj purposes.
Regarding speed, on my ZFS setup, I have 8 vdevs with data striped across those so my drive array speed far outpaces my 750/750 internet connection.