What are the best practices for configuring enterprise servers to run Storj nodes efficiently?

Hi everyone,

I’m planning to deploy Storj nodes on our enterprise servers and want to ensure they run efficiently. Can anyone share their best practices for configuring enterprise servers to get the most out of Storj nodes?

I’d appreciate any tips on hardware setup, network configuration, security measures, and performance optimization.

Thanks in advance!

Hello @edisionthomas ,

Welcome to the Forum!
I’ve passed your question on internally.
The offices are closed today for the 4th of July holiday so there may be a bit of a delay in getting you a detailed answer.
Thank you for your patience.

2 Likes

Hello @edisionthomas ,
Welcome to the forum!

The hardware doesn’t matter too much, it’s not mining, the usage depends on the customers, not hardware or software, except edge cases like using Windows VM on VMWare on the Linux host, using a network filesystems (NFS, SMB/CIFS, etc.), or using NTFS under Linux, SMR disks or USB drives without an external power supply, etc.
The hardware recommendations you can see there:

In general - VMs are highly not recommended, if you have a choice, use Linux instead of Windows, if you forced to use Windows, then also do not use VMs, use the bare metal instead.
Use 1 node per 1 disk, each node should have an unique generated identity (clones will be disqualified as soon as you bring them online), each signed with an own authorization token, each uses at least a one disk.

RAID is possible to use, if you have it, however, there would be applied the same rule - 1 node per disk/volume/pool/vdev/etc. This is related to the fact that several nodes will compete for the same limited resource - IOPS, it’s also against Supplier Terms & Conditions.
In general the RAID is not needed and not rewarded, you also usually will have a less total capacity, than if you would use disks independently.
See

For a filesystem choice you may read these topics:

It is not recommended to create all nodes at once, because all nodes behind the same /24 subnet of public IPs will share the traffic, the amount of ingress would be the same as if there was only one node: we treat all nodes behind the same /24 subnet of public IPs as a one big node for uploads and as a separate ones for downloads, repair and audit traffic: we want to be decentralized as much as possible.
Each new node must be vetted. The vetting node can receive only 1%-3% of the uploads until got vetted. To be vetted on one satellite it should pass 100 audits from it. For the one node in the /24 subnet of public IPs it should take at least a month (on Saltlake Satellite it could be faster due to performed load tests). Several such nodes in the same /24 subnet of public IPs could take more time to be vetted, because each of them would have less data and less audits due to amount of stored data.
So, it’s recommended to run the next node only when the previous one almost full or at least vetted.

1 Like

Your link does not point to an actual product, simply to the Lenovo server page.
I would advise against buying new hardware for multiple reasons.
What unused hardware do you currently own?

1 Like