SNO Capacity Planning

Question for the SNO’s

Let’s assume the data flowing in recently is consistent and your drives get filled.

Do you plan on expanding your capacity and/or number of nodes?

In other words, are you going to just let your existing drives fill and not expand further, or do you plan to expand as they fill up?

Consider that existing customers may be using all the space you currently offer. And we have customers that could triple the volume of data coming in, and over time more than that.

There is no right or wrong answer here. I just want to get a feeling from the majority of SNO’s whether they plan to expand as data usage grows or if they plan to just maintain with what they currently have.


For every drive I fill I’ll add another 20TB refurb HDD. But still nowhere near filling the first one. I don’t intend to get access to any more IPs though… so I guess I’ll keep parking the full nodes behind the same IP and letting them idle.


Send a poll in email to each sno. Here we are just a small percentage reading and answering.
Me, I will expand if the occupation is constant and I can buy more hardware with money made from Storj. Untill now I used my own money and filled all the spots.
I started this as a second income source for long term, so if the money are good, I will expand to make more.
But I’m not skilled to run data centers and I don’t have the physical space for big servers. Eventualy I will stop adding more systems and drives. My way to go is with Synology NAS-es.
Curently I have 288TB allocated and with 22TB drives I could add like 315TB. So 600TB could be the limit for me. I could go with 24TB drives also, depending on the prices.


Just ordered :

  • 27 x Seagate ST24000NM007H - X24 24TB SAS 512e/4Kn
  • 48 x 64GB LRDIMM ECC DDR4-2400 - upgrading from 512 GB → 3072 GB RAM


I will Expand, today ordered 1Gb connection to can better maintain. My 518 TB of space.


Please be careful with that assumption. Soon the TTL will kick in and the grow rate will be a lot lower than measured up until now. Do not assume the grow rate will stay as high as it was for the last weeks.

Yes I actually do. I will need about 2 months to test out a Pi5 system that I could install in 2-4 remote locations. I might be able to take a few shortcuts here and there and maybe speed up the process but that would require some commitments from the company to justify the additional risks.


If traffic keeps flowing like it does now, for sure. I just started but I don’t mind ordering some more HDD’s if I can make that amount back in a year or so.


I’m planning to add drives as the existing ones get filled, keeping in mind not to go too crazy until we actually get a signed client on and not just testing data.


Why so much RAM? It doesn’t look like you’re using what you got? Is this part of a plan to have a massive ZFS ARC?

It’s for testing - to see if it will improve speed while running retain and deleting pieces.


A few months ago I expanded with the expectation my setup would be “filled” in perhaps 2-3 years, and so far that has been cut short to about 6-7 months with the recent ingress. I have been eyeing on refurbished drive deals on both online and local markets to expand.

However I sincerely hope improvements can be made to the garbage collection process to reduce the amount of “wasted” space. I’ve already seen on Gerrit some efforts to improve this speed so that we can have more frequent collection.


5 posts were split to a new topic: Re-implement RAID0/MergeFS in storagenode

I didn’t thought we will end up runing out of wall sockets for nodes, but if it turns out that you made a real prediction, than… :sweat_smile:

@Knowledge Is this a SNOxSTORJ question? (I’m not really sure what’s supposed to go in there?)

I will keep expanding for as long as they fill up. But I am aware that once TTL kicks in I may well reach an equilibrium (I’m not getting more than about 150Mbit throughput on my 1Gbps service, for some reason)

EDIT: BUT… you do really need to look at the whole databases and file walkers situation.


I’m cautiously expanding already. At the moment there is way too much uncollected garbage and trash for me to expand too fast. That space is going to free up at some point. Additionally we haven’t had the TTL kick in yet. So yeah, most of my IP’s no longer have nodes with free space. But I only plan to expand when they all fill up. For now. I might expand more liberally when things are a little more clear and settled.

I’m also really looking forward to those 30TB HDD’s. My drive bays are full and I don’t really feel like adding more. So I’m removing smaller HDD’s every time I add a big one and I’d rather just add the biggest available when I do.


That’s gonna be a lot of tokens! :wink:

EDIT: I am not entirely comfortable with having so much data in one HDD.
File walkers already take an inordinate amount of time and databases are getting very large as it is.

1 Like

With a 30 days TTL, could the maximum size of the SN be calculated with the internet connection speed?
Imagine a constant 500Mbps ingress (LOL) during these 30 days, it’s around 5.15TB ingress/day… 160TB per month… and after a month, data will start going to trash


I look forward to the day I need to replace small-HDDs with large-HDDs (because I’m out of bays). While I still have bays all I can do is not use anything smaller than 20TB (even though used 8TBs are cheap) and try not to make my future problems worse :wink:

May be i understand block size wrong but even 16 TB HDD have minumum 8k blok size, this big HDDs can endup with very big Bubles as storj use very small files.

1 Like