Storagenodes for high performance usages - tiering and tiers

Hello Storj

Today i would like to suggest you some services, about to offer a new bunch of functions regarding different use cases :

My entire suggestion requires a storagenode evaluation to determine internet connection speed and maximum throughput in read / write on storage.

1. Performance cloud storage
Cold object storage representing the actual Storj network is well, It permits to run multiple cases of data instances into the cloud ! Backup, archiving or data flows with a minimum of performance are most common cases. My suggestion is to offer a performance service, for example at more than 60MB/s into read operation for customer uses.
1.bis Performance cloud storage modulation into region and custom erasure coding factors
Performance cloud storage should be offered with custom erasure coding level possibility (number of times the data is replicated) and should be offered also with the ability to be set on a given region (area of the world). (1.bis would permit storage for less price at “customer risk” or test purposes).
2. Rename actual Storj V3 network “Archive tier”
This corresponds actually to archiving cases, or backups with long retention and no important RTO constraint.
3. Normal Tier
Like performance tier, create a “normal” tier with maybe 40MB/s possibility.
4. Setup Tiering into code and Storage Lifecycle Policies
Permit to frequently accessed data to move up into performance tiers with auto-tiering when object is accessed, permit to objects to move down into archiving or normal tier when not accessed. For Storage lifecycle policies, permit the customer to choose after a period to move their data on another tier programmatically (ex move to performance to normal then archive tier).

Thanks for keeping us informed in case of evolution possibilities next to this topic.

So is the change going to be finding a way to slow down speeds for certain customers? Because 500MB/s and higher is already achievable for those who have the Internet speeds that can handle it.

Since customers download from SNOs directly (after finding out their IPs from a Satellite) I think that means you’d need to add a way for independent SNOs to “work together” to throttle as a group? And SNOs are already competing to win transfer races: would that mean instead of faster SNOs earning a bit more … they’d no longer benefit from being fast… as they’d be “held back” which would put slower SNOs on equal footing for payouts?

Geofencing is already possible. Afaik.

Also 25GB free per account.

I do not know about the status of “pieces of the erasure coding” setup, but Storj does not use “replication”. and similar suggestions were made.

this is partialy realised at Node level.(some nodes have ssd/nvme/ram cache for frequent used pieces.)
V3 Network offers all tiers in one (i think setting the piece-count is already possible)

There is an additional SOC2 level (for mostly buerocratic purposes?)

1 Like

Hello @Eioz,
Welcome back!

How do you suggest to measure the actual speed for the node which introduce itself as fast (I suppose in the config)?

We do not use replication,

The geofence feature is already available for paying customers.

The moving data by the repair workers is already implemented, but it’s expensive. So it’s not a default option.