3 VERY LIMITED "GET GOOD SLEEP" STORAGE PLANS FOR ALPHA TESTING

risharde

Hi everyone, I spent a few months working on this storage project which is still in alpha but I really need basic testing just to ensure I'm not the only one that thinks it's stable (and then I can at least call it beta).

I'm looking for at least 3 reputable persons on the forum (but preferrably not providers for obvious reasons of competition at this point) who would test the service. Reputable should be people with at least 1 year of existence here or on OGF. At least 50 non-spammy comments would be nice.

This service will be "free" for AT LEAST A MONTH OR MORE (but obviously I need feedback as to whether it's stable or one experiences any quirks - so that's the catch). As a great thank you if issues are found, I'd really try to extend that time as a thank you to the testers. There might be times where the panel could go down when I do updates but that does not affect the SFTP access in any way.

If all goes well, I want to develop a second phase which will add some super cool features (since right now it's just SFTP access).

Anyways, enough talking, here's what you get:

100 GB SFTP Storage (how boring) but...
1. This 100 GB of SFTP is on RAID 1 and is located in Canada (250 mbit unlimited: OVH).
2. The storage is NEVER OVERSOLD (which hopefully means access times should rarely if ever be affected by oversold storage). At this point, looking at a max of 20 users if it's 100 GB blocks, less if it's portioned in bigger blocks.
3. The 100 GB is replicated to another dedicated server (which also is on RAID 1) in a different geographical region (in this case London - OVH) periodically (more details can be provided once you're selected to test) and will be adjusted based on feedback and testing.
4. A custom built "technical" storage panel that doesn't just show you just disk space. I'm nearly sure you've never seen ANYTHING like this (even though it's not necessarily aesthetically pleasing) but when I say ANYTHING - let's just say you literally know what's going on with the server and storage details. Hint: IO WAIT, TEMP, UPTIME AND DEEPER DETAILS. This is a super important part that I hope makes the service worth the planned price (if it ever reaches public release).
5. And most importantly, "you can sleep good at night" since you don't have to guess what's going on about your storage, it's stored in 2 different geographical reasons and you in theory have access to both servers in the rare event one of the servers go down (both accessible via SFTP with the same login details).

This package when I did the numbers would be around $4.50 a month (assuming 50 cents out of this goes to the payment gateway) if I ever reach the point of releasing to the public.

I understand this is totally expensive but you'll get more details as I said other than "cloud" storage where you just totally rely on the word of the company. I hope the value of having to not worry while sleeping would be worth it?

I have personal interest in this project since I plan to also be my own "customer" - I really want to sleep better at night knowing my data is safe from loss (to a higher degree). This in theory is still not considered a backup but I think it's more robust than guessing and hoping people are doing your storage right for you.

So please PM if you're interested (it's a manual set up and would require you to register an account at ONDDNS.com since the panel uses my own SSO technology).

Thanks again at least for reading even if you don't decide to PM in.

Additional rules: You agree to not screenshot, share further details with others at this point until launch IF all goes well - launch would be delayed by a few months well IF I get to 2nd phase of development.

risharde

1 account issued, 2 more remaining up for grabs!

pikachu

@risharde said: Hint: IO WAIT, TEMP, UPTIME AND DEEPER DETAILS. This is an super important part that I hope makes the service worth the suggested price.

The real question is - why should I have to care about any of this? Ideally, I'd like something where I can write any volume of data and it automatically takes care of such concerns. See Amazon S3 or even a smaller object storage provider, as an example.

risharde

@stevewatson301 said:

@risharde said: Hint: IO WAIT, TEMP, UPTIME AND DEEPER DETAILS. This is an super important part that I hope makes the service worth the suggested price.

The real question is - why should I have to care about any of this? Ideally, I'd like something where I can write any volume of data and it automatically takes care of such concerns. See Amazon S3 or even a smaller object storage provider, as an example.

@stevewatson301 That's a good question and that really depends on the individual. I was motivated to build this from my experience with storage on hosts (never used S3 nor would I unless that had "unlimited" bw and SFTP which is easy to mount on SSH as far as I've been seeing people requesting on storage) and I wanted more insight on what was happening on a storage node. If you've ever asked for details on a support ticket and gotten "we're working on it" - I'd prefer more details on before hand if something is up. For the end user, being able to keep watch on whether there's an error on a drive instead of a panel saying "online" when it's not is hopefully going to be useful but again that's just my needs. It's not going to be for everyone if they don't care about those details.

This might also be more for the techies who want to know or have anxiety worrying about if their data is on drives / configurations that are in good status.

I use for example google drive (for critical data) which is a "cloud service" and should be safe but I don't know what that safety involves and if that safety can fail - I definitely wouldn't put all my trust in the service even if they said it was bomb proof. I'm not saying my system can't fail but I'm attempting to drop those odds with raid 1 mirroring and geo replication built it so that user doesn't even have to do that part - they just mount their drive or directly SFTP and they're on their way and in the process offering to show the server details for every "customer" to also ease their minds and in the process keeping me in check as well as an upcoming provider if this works out - since the user gets to see if a drive in an array is down for example or not.

I hope I answered at least my motivation, I'm not a sales person so I'm just being as honest as I can here and most of my services will be based on that IF it ever goes public.

cybertech

just curious what raid will it be?

james50a

i like storage stuff but tbh in this case i feel like maybe i missing where the value is? I guess kinda the main concern here is why would I pay 10x the rate of something like a hetzner storage box (3.45eu 1tb) for the same sftp access with known raid setup? Apologies if i missed something obvious in your initial post.

risharde

@cybertech said:
just curious what raid will it be?

RAID 1 on both servers. I could cut corners but that would make me lose sleep at night.

@james50a said:
i like storage stuff but tbh in this case i feel like maybe i missing where the value is? I guess kinda the main concern here is why would I pay 10x the rate of something like a hetzner storage box (3.45eu 1tb) for the same sftp access with known raid setup? Apologies if i missed something obvious in your initial post.

Np no need to apologise, you weren't mean making your point.

A few things: you don't know how hetzner is storing your data (not that they are liars but just saying). My control panel shows both independent servers and allows you to log into both to see your replicated data. I think hetzner has a connection limit on their plans? There isn't a hard limit here (that may change to avoid abuse but I think it will be manageable as I'm not paying hundreds on these dual servers. Storage - I am totally beaten on, that's expected. If you sleep good at night with your current storage provider, it isn't for you but if you don't, then it's probably for you?

risharde

I want to mention that so far in building this, I do have other target audiences I am seeking to provide this storage for (I would love to match pricing for low end users but that's super hard to do). I really hope the 2nd phase of development which is the value added features of the panel will be what is going to add more value than it just being an sftp drive.

bikegremlin

My thinking - if it helps you with any ideas:

Even with raid 1, I'd still use another copy in order to sleep well.
Hetzner has been reliable enough - not a single problem so far (knock on wood).
Similar goes for Backblaze B2.

If they start requiring monthly (or even annual) re-uploads because they crash, I'd start looking for an alternative.

As it is now: why would I go with another storage instead of either of those two?
That's the selling question.

Encryption and zero-knowledge are the first answer that pops to mind. At least for the not-very-paranoid customers who would still encrypt before uploading either way (for those it's not a pro I suppose).

RAID 1?
I consider it to be more of an uptime thing (not saying I'm not wrong, but that's my point of view).
I'd still need a second copy - either on the "cloud" or in a remote physical location.

deank

No offense, but "testing", "storage", "alpha". "good sleep" can never be in the same sentence.

When those words are mixed in a blender, the only thing that comes out of it is manly tears.

Boogeyman

@risharde said: So please PM if you're interested (it's a manual set up and would require you to register an account at ONDDNS.com since the panel uses my own SSO technology).

I couldn't access onddns.com. Btw do you use https://github.com/keycloak/keycloak for SSO in onddns.com?

risharde

@bikegremlin said:
My thinking - if it helps you with any ideas:

Even with raid 1, I'd still use another copy in order to sleep well.
Hetzner has been reliable enough - not a single problem so far (knock on wood).
Similar goes for Backblaze B2.

If they start requiring monthly (or even annual) re-uploads because they crash, I'd start looking for an alternative.

As it is now: why would I go with another storage instead of either of those two?
That's the selling question.

Encryption and zero-knowledge are the first answer that pops to mind. At least for the not-very-paranoid customers who would still encrypt before uploading either way (for those it's not a pro I suppose).

RAID 1?
I consider it to be more of an uptime thing (not saying I'm not wrong, but that's my point of view).
I'd still need a second copy - either on the "cloud" or in a remote physical location.

Totally understand, the 2nd server has a copy in a different geographic region but it's not an archival 'backup' and thus indeed user backups I suppose are still a responsibility of the user but to be honest, because the replication to the backup is not instant, if for example there was an issue with the primary server, it could be shut down, allowing users to pull from the secondary server in the event of a catastrophic failure which I have not experienced ever knock on wood

Uptime is a good point and to some degree mitigated by the 2 servers being in different locations (this is good for access but not writing) since there is a one directional replication that happens from primary to secondary. The bidirectional I feel is less safe but didn't test it or think users would be interested in that. This would probably be more for the programmers to integrate code to switch to the secondary server (and while I can do this for the user, it wasn't part of the plan since I want to keep the service simple to be able to provide quick support)

There's a few things my service is adding (whether it's valuable or not is all dependent on the user's point of view)

1. More transparency at the hardware level
2. I'm really trying to build this with being aware of how brutal things can get in the lowend forums when things go wrong so I am making the effort here to ensure stability comes as a high priority. Also to some degree that I will try in me personal capacity to keep my name clear from doing the wrong thing like lying about what I am offering etc (my personal name and reputation is selfishly important for me to uphold)
3. And back to product, more 'dedicated' like resources and at this point no major hard limits (greater tolerance support wise for users, I hope)
4. Replicated server with data in a different geographical location (though getting these servers at this price point has been difficult)
5. The value added features I intend to add with regard to media (hint) might be tempting (hopefully)

risharde

@deank said:
No offense, but "testing", "storage", "alpha". "good sleep" can never be in the same sentence.

When those words are mixed in a blender, the only thing that comes out of it is manly tears.

Intent and result are two different things, I would have had to write much more pages to explain it. What I have right now is the start and not the fully end product since I am appreciative of ensuring the main basic part works. I do appreciate your point though - never been good at articulating everything but I will try to make an extra effort to be more clear if I can.

deank

Right..., as some troll would say, you must be fun at parties.

risharde

@Boogeyman said:

@risharde said: So please PM if you're interested (it's a manual set up and would require you to register an account at ONDDNS.com since the panel uses my own SSO technology).

I couldn't access onddns.com. Btw do you use https://github.com/keycloak/keycloak for SSO in onddns.com?

Sorry that you cannot access onddns.com, strange I will look into it but if you have any details on your side - for example if you literally cannot resolve the hostnsme, I can access look into it further. I can currently access it using 8.8.8.8 and 8.8.4.4. By default http does not work just in case you didn't try https (and remove the www if you by chance used that). Please let me know if you have the time.

Not using keycloak - totally coded something less interesting than that lol

Thanks for the questions, sorry again that you cannot access it!

risharde

@deank said:
Right..., as some troll would say, you must be fun at parties.

Lol I'm not fun at parties for the record lol

msatt

Just a suggestion - I am based in the UK so having this as a secondary is counter productive i.e. main is in Canada etc.
If you are considering Europe then the master / slave relationship needs a bit more thought,
Good luck and have fun.

pikachu

@risharde said: you don't know how hetzner is storing your data (not that they are liars but just saying). My control panel shows both independent servers and allows you to log into both to see your replicated data.

There's no reason to trust you either; how do we know the two servers aren't actually the same? Hostnames etc. can be faked, and one could always write a custom SSH server. (This is trivial given the abundance of SSH libraries these days.)

@risharde said: greater tolerance support wise for users

Wait till you see some real PMSing users, your patience would be worn down quickly.

risharde

@msatt said:
Just a suggestion - I am based in the UK so having this as a secondary is counter productive i.e. main is in Canada etc.
If you are considering Europe then the master / slave relationship needs a bit more thought,
Good luck and have fun.

Thanks, that's a super great point! I totally thought about this! I will consider such a combination once the demand is there for the service which is going to be hard to determine without a launch but rest assured, if it's there it will happen where the primary will be in the UK.

I will iterate another set up where the primary maybe in another European country for others closer to countries other than the UK as well.

That's will be after 2nd phase of feature development when I add some cool functional features to make it more than just sftp.

cybertech

is raid 1 sufficient IOPS for multi user scenario?

especially since the panel is showing IOWait which can be pretty high during usage.

deank

I would rather not have RAID actually.
It doesn't really help. All it does it give you downtime.

risharde

@stevewatson301 said:

@risharde said: you don't know how hetzner is storing your data (not that they are liars but just saying). My control panel shows both independent servers and allows you to log into both to see your replicated data.

There's no reason to trust you either; how do we know the two servers aren't actually the same? Hostnames etc. can be faked, and one could always write a custom SSH server. (This is trivial given the abundance of SSH libraries these days.)

@risharde said: greater tolerance support wise for users

Wait till you see some real PMSing users, your patience would be worn down quickly.

That trust will hopefully be built over time (stressing greatly on hopefully). I've been on ogf for years now and many of us put our trust in hosts we've never seen by people who recently joined and have little to no footprint - many times not even from users with their real names and not even with real photos of themselves - I use my real name, email etc. My photo is real. These might seem to be small things, some might even say it's stupid for security reasons but I am as real here as I am real in my physical life. With that being said, I don't want people to trust me instantly, the panel provides info to the customers which I hope people will eventually trust. That would be a combination of seeing what the panel says, testing network speeds, write speeds, overall uptime, logging into the secondary servers to see the data is replicated data as well. Obviously some things are going to be super hard to prove for example I don't know of another way to prove the locations of servers other than giving the IPs and even that can report in wrong locations but again, I plan to be my own customer since I have other plans and services I want to build. Even at 4.50, I would have to fill most of the server slots (20 at 100 gb) to get any kind of profit of which I could probably buy one hamburger for the month if I fill one server (lol). It isn't primarily about profit at this point (and my capital investment is from my own salary), it's about me being able to sleep at night and sharing such a service with those who may be on my level of anxiousness. I don't even want to expand fast because I want to be able to ensure support is manageable though if all goes well, the idea is that the system itself is robust enough to not require hundreds of hours of support (due to stable software and controlled customer intake). Right now because it's sftp, don't see much support being required as is. But either way, even that part I have a unique idea for which I hope to bring in!

risharde

@cybertech said:
is raid 1 sufficient IOPS for multi user scenario?

especially since the panel is showing IOWait which can be pretty high during usage.

Definitely a great question! I think this would all depend on the users and how they are using it. By having a relatively small amount of dedicated users (price goes up as you can see) but hopefully performance does not suffer greatly. I am far from being a provider with that level of experience but I don't forsee at the moment that level of abuse that causes the storage to cripple. If a bad actor comes in to play, I will set a limit or refund them if it potentially affects others (and at most 20 customers vs a hundred since that's the max I plan to put on a dedi is 20 at this point).

I plan for this to be very controlled. I don't plan to rake in thousands of dollars of profits either.

risharde

@deank said:
I would rather not have RAID actually.
It doesn't really help. All it does it give you downtime.

I can see your point but what if your single drive fails, that would cause downtime too? Not arguing with you, there's just a few scenarios other than raid bring an issue. Also downtime can theoretically be mitigated to some degree if one switches to the secondary server for retrieval. I don't recommend using the secondary server for writing because of the one way replication. Again I can rethink enabling two way sync but I don't know if users would prefer this or not.

deank

Since I always have backups, I'd rather have as short downtime as possible so that I can begin the restoration.

For those who do not have the habit of maintaining backups, I don't know what to say. This is year 2022 where everything is becoming digital. Backing digits ain't hard. It is in fact far easier than making backups of physical data.

risharde

@deank said:
Since I always have backups, I'd rather have as short downtime as possible so that I can begin the restoration.

For those who do not have the habit of maintaining backups, I don't know what to say. This is year 2022 where everything is becoming digital. Backing digits ain't hard. It is in fact far easier than making backups of physical data.

I would love to add backups to make the service line absolutely a set and forget type of service but I feel like the 4.50 is already high reaching and I'm trying to cater for the low end market if I can as well but pretty sure a lot of people won't bite due to the price sensitivity.

But really I would love that one doesn't even have to reach the point of backups (maybe wishful thinking).
I could manually point the primary server hostname to the secondary server to allow reads and if users are okay - allow write access making the secondary become the primary in the event that fixing the primary server requires major downtime if that makes sense.

There's a catch there because in theory, with one server down out of the two, it's obviously a point of anxiety but still again, much less than not knowing what's happening.

deank

I never purchase a backup service from the same provider.
That ain't a backup.

A real backup is a copy of what is essential at a different location.

risharde

@deank said:
I never purchase a backup service from the same provider.
That ain't a backup.

A real backup is a copy of what is essential at a different location.

TRUE, that's wise! But if I were to add a backup (to be added within the service - which at this point seems to be too much of an automation hassle) - I would probably indeed be doing that with another provider and that part might not be so much having the requirement of having RAID (so many options it would ruin me trying to keep it simple and manageable)

deank

You, as a provider, are free to provide all-in-one package for extra profit.

But it is also entirely possible to provide a very lean service without promising too much.

An eternal issue with providers around low end market segment is the desire to grow bigger. The ultimate end result of that is a heavy fall. Perhaps, try not to grow and just provide a lean service.
That's what Fran does, doesn't it?

ralf

100GB for $4.50 a month sounds expensive. I suspect most people would rather have multiple backups on different cheaper providers than one backup on an expensive provider who promises that it's RAID under the hood, because that's only addressing one point of failure - the disk, whereas proper multiple backup locations can give you redundancy against network failures, geopolitical issues, etc as well as disk failure.

risharde

@ralf said:
100GB for $4.50 a month sounds expensive. I suspect most people would rather have multiple backups on different cheaper providers than one backup on an expensive provider who promises that it's RAID under the hood, because that's only addressing one point of failure - the disk, whereas proper multiple backup locations can give you redundancy against network failures, geopolitical issues, etc as well as disk failure.

Totally feel you here, I don't know how other providers could be charging so cheap without cutting corners or overselling (not that overselling is always bad). However most of the other providers do not usually advertise the offsite (in my case secondary server) with the same access as the primary. The price essentially doubles because I'm using 2 servers as-is BUT the customer is able to log into any of the two servers. There will be much more value added when I get to the 2nd phase which are some unique features related to how a customer can use the storage to their advantage for things other than just SFTP transfers. More on that when at least I reach alpha there as well (don't want to jinx the project)

P.S I must mention here, as features increase on any services I make public - I plan to give out complimentary services IF all goes well. Again this will not entice everyone but being a person who is also price sensitive - I'm hoping to reach a point where that price will be a wonder. Lots of hopes and limited time to get it rolled out. Just for example, being a customer of the storage plan would also add higher thresholds for the ONDDNS service. I also have a grid page project which is in alpha which I plan to bundle in as well. So you'd pay for storage but get access to 2 additional services (or maybe more if time allows it).

I'm also looking at how I can make this in such a way that it isn't going to be a one man show (without continuity). Lots to think and do.

deank

100GB for 4.5 is indeed expensive.

I mean, I pay 2 CAD for 100GB from google.

chimichurri

@risharde said:
I don't know how other providers could be charging so cheap without cutting corners or overselling (not that overselling is always bad).

Off the top of my head:

• Using DCs with cheap electricity
• Owned (and not rented) hardware/IPs
• Scale
• Automation

risharde

@chimichurri said:

@risharde said:
I don't know how other providers could be charging so cheap without cutting corners or overselling (not that overselling is always bad).

Off the top of my head:

• Cheap electricity
• Owned (and not rented) hardware/IPs
• Scale
• Automation

Probably but I have a hard time believing cheap electricity drops it much further than OVH which is rather cheap already as is. Also what I think a lot of storage providers are doing (yes I'm guessing but hopefully intelligently) is adding HUGE amounts of disks to single servers - this allows a huge number of users but when things go wrong - they go wrong big - affecting a higher number of users when something goes wrong. It's the price one is willing to pay to have customers and how much one is willing to put their reputation on the line AND how much one is willing to handle the stress of users being angry (which they would rightfully be) when things go wrong.

I'm heading in the opposite direction IF I can avoid it. Worse case scenario is that its not a product people want (which would be saddening BUT understandable) but I'm hoping that's not the case and it's worth the sleep - it has been for me at least (and I have been price sensitive but again after some of the stuff I've experienced and read - the 4.50 is a hard sell but not when I compare it to my sleep and reduction in anxiety)

I agree with scale and automation for sure btw! Wasn't meant to be a refute to what you mentioned!

risharde

@deank said:
100GB for 4.5 is indeed expensive.

I mean, I pay 2 CAD for 100GB from google.

I'm aware - I can possibly drop that but I'd have to use the OS drives as well - not too comfortable with that option. Plus that would add another 15 users to the mix which I don't know I really want. Again, I really want to avoid people having issues and being mad with me - very few are going to say hey it's cheap and still not expect it to be up most of the time)

chimichurri

I think OVH have checked all of these four boxes (in terms of electricity, esp. their cheapest offers i.e. Kimsufi are AFAIR located right next to the local power plant, hydro in Quebec and nuclear in France). And yes, I think a regular storage node has a lot of drives attached to it, example: https://www.hetzner.com/sb?country=ot&search=1722930

I have to admit, I'm also skeptical overall. I don't really care about temperature data, and no offense, but just because you're using your real name and IRL picture doesn't mean I'd automatically trust you with my data any more than any of the established providers... whom BTW I don't trust 100% either - ideally, I try to go with the 3-2-1 rule of backup, or at the very least, to keep two redundant copies at separate locations/companies.

And with this being LES, price-wise, 4.50 a month is indeed a hard sell, lol, aside from some already sold out BF offers, other than the already mentioned Hetzner, here's another deal you'd be competing with: https://clients.servarica.com/store/black-friday-2021/polar-bear-storage-offer - 2TBs of RAIDed storage for 4 dollars a month (admittedly, paid yearly)

Sorry if I sound like a buzzkill, but given all what has been pointed out until now (mostly by others on p1), unless you find a unique selling point, I'd really think this through if I were you - I'm pretty sure that given your lack of reputation compared to the aforementioned likes of Hetzner, B2 and S3, as well as the hefty price tag, it won't turn out to be a product that people would want - ergo, I can't imagine you making any kind of decent money with this.. :'(

deank

Well, it does not matter in the end

because the end is nigh anyway.

pikachu

@chimichurri said: I'm pretty sure that given your lack of reputation compared to the aforementioned likes of Hetzner, B2 and S3

Not to pile any further, but the angle OP has been pushing, namely that Google and Amazon are defrauding their customers by pulling off a cociu style basement host operation is honestly a bit much.

@risharde said: 4.50 is a hard sell

And not just for lowend folks. S3 and GCS has a lower price than your offering $0.023/GB, and I'm yet to hear about a single incident of objects being lost from these services.

@risharde said: Also what I think a lot of storage providers are doing (yes I'm guessing but hopefully intelligently) is adding HUGE amounts of disks to single servers

Object storage providers forego RAID as a concept entirely and directly implement Reed-Solomon on commodity hardware with all the distributed systems goodness (think Paxos, quorums etc.) to implement highly scalable storage systems.

image_host

I made pricing for my storage site with this setup.

1) RAID6
2) Same data in 3 DC
3) 2.5USD/100GB

With unlimited BW.

deank

I tell ya one thing.

The only storage I trust is my own that sits in my basement in a form of 1U in a rack which no longer has spindle drives anymore but two 2TB NvMe drives.

My 100GB google drive is a family drive they share stuff online.

risharde

@stevewatson301 said: Not to pile any further, but the angle OP has been pushing, namely that Google and Amazon are defrauding their customers by pulling off a cociu style basement host operation is honestly a bit much.

That's not what I said and why are you continuously speaking about object storage? My intended service isn't to provide object storage nor be the next Google or Amazon - they work on massively different principles conceptually and even financially.

@image_host said:
I made pricing for my storage site with this setup.

1) RAID6
2) Same data in 3 DC
3) 2.5USD/100GB

With unlimited BW.

Congrats to you, I wish you all the best!

[@chimichurri said] Sorry if I sound like a buzzkill

No worries, if it's not for you, it's not for you. If it's not for many, then it's not for many. If it's not for any, then it's not for any.
Worst case scenario is I use it for myself. Would be nice if people saw an appreciation for such a service but if not, that's fine.

@deank said:
I tell ya one thing.

The only storage I trust is my own that sits in my basement in a form of 1U in a rack which no longer has spindle drives anymore but two 2TB NvMe drives.

My 100GB google drive is a family drive they share stuff online.

I use Google Drive for critical data but I make backups but mounting Google Drive doesn't seem like a thing for me to do with my current applications which is another reason I began developing this plus I wanted to be in control of what was under the hood code wise.

Anyways, if I were to guess based on the comments - it's an awful idea for you guys so I'm gearing up to not have many customers but we'll see how it goes. I'm super happy that people were willing to test it out and give me feedback and also this has motivated me to work on all the rest of goodies.

deank

You can certainly do super premium storage. There is clearly a market segment there.
But we are obviously the wrong crowd to talk about.

And, unlike super-duper cheap storage, you can't go on creating premium service and expect people to embrace you. You need a backing, a known entity, to vouch for you (translation: bribe).

lentro

@stevewatson301 said: Object storage providers forego RAID as a concept entirely and directly implement Reed-Solomon on commodity hardware with all the distributed systems goodness (think Paxos, quorums etc.) to implement highly scalable storage systems.

Backblaze publishes how reliable they are very clearly, and it's probably more reliable than 2 machines: https://www.backblaze.com/blog/cloud-storage-durability/

They have 12 nines of durability and only claim 11 nines lmao

Not sure why ppl wouldn't just upload their archival storage to like 2 cloud services for like $10/TB/month in storage (e.g. Backblaze B2 + Storj) and call it a day.

CamoYoshi

While I like your idea of coming up with unique features I'm not sure it will justify the cost to the customer at the end of the day. Only way I could think of making yourself stand out would be something like shared unmetered 10Gbps and storage fast enough to handle ingress/egress at those speeds. Unfortunately in the realm of cloud storage it's a constant race to the bottom, so starting the ball rolling will be really hard. I do wish you the best of luck, genuinely, and hope you find the customer base to make this a real thing as I love the idea of being transparent with regards to systems statistics.

lentro

@risharde said: My intended service isn't to provide object storage

Can you elaborate a bit? If this is for long-term archival storage, don't the two kinda serve the same purpose?

@risharde said: it's an awful idea for you guys so I'm gearing up to not have many customers but we'll see how it goes

Don't listen to the haters! if the product is good, if you have a gut feeling that many people will use it, go for it. But I will share that when we launched TensorDock on LET/LES, we issued around $2.5k in credits, from the students to the startups. We got around $5 in real revenue I would guess. But it was the LET users who then told their friends, who then put it in Discord servers, that helped us grow in the long term. We also have our own sales that focuses on larger deals, so I'm not sure how much revenue is attributable to LET/LES, but I'd guess it's probably worth more than a few thousand dollars. Just have to not piss off too many people on this forum

risharde

@lentro said:

@stevewatson301 said: Object storage providers forego RAID as a concept entirely and directly implement Reed-Solomon on commodity hardware with all the distributed systems goodness (think Paxos, quorums etc.) to implement highly scalable storage systems.

Backblaze publishes how reliable they are very clearly, and it's probably more reliable than 2 machines: https://www.backblaze.com/blog/cloud-storage-durability/

They have 12 nines of durability and only claim 11 nines lmao

Not sure why ppl wouldn't just upload their archival storage to like 2 cloud services for like $10/TB/month in storage (e.g. Backblaze B2 + Storj) and call it a day.

I use backblaze but not the block storage for backups - that leaves a lot to be desired BUT it's "unlimited" - mostly for some personal data that is also backed up to hard drives.

@CamoYoshi said:
While I like your idea of coming up with unique features I'm not sure it will justify the cost to the customer at the end of the day. Only way I could think of making yourself stand out would be something like shared unmetered 10Gbps and storage fast enough to handle ingress/egress at those speeds. Unfortunately in the realm of cloud storage it's a constant race to the bottom, so starting the ball rolling will be really hard. I do wish you the best of luck, genuinely, and hope you find the customer base to make this a real thing as I love the idea of being transparent with regards to systems statistics.

Thanks, totally wish I could get 10G links, that would indeed hopefully make the intended deal more tempting! As for stats - there's much more important stats in the panel that I have not made public and indeed I think you grasped why I built it from a user perspective. I really hope I can get things to a release state just even to share it without having be secretive about the details - that's my next motivational factor!

risharde

@lentro said:

@risharde said: My intended service isn't to provide object storage

Can you elaborate a bit? If this is for long-term archival storage, don't the two kinda serve the same purpose?

@risharde said: it's an awful idea for you guys so I'm gearing up to not have many customers but we'll see how it goes

Don't listen to the haters! if the product is good, if you have a gut feeling that many people will use it, go for it. But I will share that when we launched TensorDock on LET/LES, we issued around $2.5k in credits, from the students to the startups. We got around $5 in real revenue I would guess. But it was the LET users who then told their friends, who then put it in Discord servers, that helped us grow in the long term. We also have our own sales that focuses on larger deals, so I'm not sure how much revenue is attributable to LET/LES, but I'd guess it's probably worth more than a few thousand dollars. Just have to not piss off too many people on this forum

Wow that's a great experience and super insightful! Super happy it worked out for you considering the investment you had to put out in those credits as well! Storage is definitely a saturated type of market AND also because storage can mean so many things. To answer your specific question - it's SFTP and yes I suppose it could act like object storage especially if you just consider object storage to be an extra layer of how the underlying storage is managed but I meant it wasn't intended for just backups. I'm assuming there's going to be a few folks who will mount it as a drive to extend their current servers and might want to use it a little more frequently for things other than backups.

The value added part of it would also mean the "public" section would be available to whomever the owner wants to share their data with. I plan to make it hopefully easy to get data in. Data being very loosely termed here.

deank

If people ask you to elaborate, it's doomed to fail.

That's my take on it.

hornet

It seems like there are several people in the thread who are determined to poke holes in your plan. You have been courteous and professional in your replies, which speaks well to your business comportment. You have stated that there will be value-added services and/or premiums, and that is fine -it seems to have frustrated the trolls to not have something further to assault. You do not have to divulge those details at this point (while you are starting with phase one). Good on you. Keep up the good work & the good attitude!

risharde

@hornet said:
It seems like there are several people in the thread who are determined to poke holes in your plan. You have been courteous and professional in your replies, which speaks well to your business comportment. You have stated that there will be value-added services and/or premiums, and that is fine -it seems to have frustrated the trolls to not have something further to assault. You do not have to divulge those details at this point (while you are starting with phase one). Good on you. Keep up the good work & the good attitude!

Thank you for your response and for being positive and encouraging, I really do appreciate it! I'm no guru at life but I've had some experience with having to please individuals as I'm sure most everyone has and it can be good sometimes difficult other times - there's just so far one can go and I think life in general requires us to take an initial chance on someone whether it's just making friends or making business partners. I am not a businessman, a sales man etc, just a programmer who loves to build things I think might br useful and as clichéd as it sounds, the greatest happiness is when I can create something others can enjoy using and of course having done that for a while, I keep the expectations low so any, and I mean any positive feedback goes a very long way. With that being said, I know how I would like to be treated by a provider and that's what I will try to do as well for anyone who will use the service if it becomes a reality (unless there is abuse involved). I'm not passionately motivated by profits, I really want to solve a problem I had and maybe I will solve this problem for others AND try my best to keep my name in good standing. Let's hope I can patch any reason holes pointed out by all those who shared their views.

P.S I don't know if some of the posters were trolling me but I didn't feel like they were not to refute your insight regarding what was said. My affinity is to allow everyone (mentally speaking) the opportunity to share their views once it doesn't get personally disrespectful (without reason).

I look forward hopefully to share some of the new features in the near future - that will hint of a launch IF I and the current testers are good with it!

Thanks to everyone who shares their insight, they will continue to be deeply weighed as I make those next steps!

deank

Don't trust me. You will thank naysayers later.