Thanks to everyone who commented in, I took a few days break away from the comments to think more about the idea. I'm convinced the idea will not work as many of you mentioned in terms of it simply being a storage alternative to much larger organizations. I'm still going to continue my second phase which is software/feature related. I have to re-iterate that I'm solving a personal problem which I was hoping to share with an expected smaller subset of the LE community (because I know it's expensive). I'm also very much aware of price sensitivity - I'm a bit surprised that many of you put your trust in the large organizations (I know they have good tech and skills but I just cannot see myself putting my trust in for example Amazon - I know they have come a long way and have become better at what they do but neither can I say I feel comfortable putting my data only on Google as well) NEVERTHELESS I APPRECIATE THE FEEDBACK SINCE IT'S A LEARNING LESSON!
The project basically started off solving my own challenge which is trust only to be shut down by others on the merit of trust as well (ironically) so I'm still motivated to solve the problem I had previously which is data loss. This week the provider I store my backups with had to go down due to node issues and it still seems to me that this project again is necessary for me - maybe one day it will be worth merit and some of you will reconsider my stand point. I've said this throughout the years actually that I have very little confidence in RAID10 which to most of the IT world seems to be heressy but say what - everybody has their own taste. I'm just really trying to sleep good at night (as funny as that sounds and as much holes poked into the implementation) - I've solved this partially for myself.
There's a next part I'm trying to solve for myself which is ease of use from another angle and I'm being very careful not to release that info yet but I'm hoping it will be relatively the exciting part to see what can be done with the storage - I know for sure others will poke holes in that as well but I'm still doing it because I really need it. I'll still likely make offers available - and if people want to come on board - by all means that would be great - even if it's just enough to fill one batch of nodes - it would still be awesome. IF no one use it, bummer, but not the end of the world. I love building things and again the most enjoyable part (other than actually a profit - while small) is actually people using and liking what I build - can't say I've achieved that here as yet but I'll stay optimistic.
Thanks to all the testers so far - you pretty much have storage space indefinitely for free and also thanks to all those who commented in with feedback (even if I saw it as negative - it's still probably realistic so I appreciate it!)
Will keep everyone updated as things hopefully progress!
I noticed this discussion today only, it was interesting read. @risharde does seems to be passionate and defending his idea was great. But later on accepting that it may not work was kind of smart.
Whatever you are doing it, keep doing just try to think what problem are you trying to solve which is not solved by others already or atleast you can do better than them. Even if it fails, you will still gain a lot in terms of knowledge and experience which you might put to good use elsewhere in future.
Hi everyone! I assure you, I have not been sleeping but I have been a bit quiet as I worked through making this make sense. My primary goal was to build an environment where data was replicated to two physical servers - something that seems to trash the whole idea of being 'cheap' or 'low end'. Again, I hope I considered each of your points correctly and I am going to downgrade my idea back to a single server with RAID. This will half the cost instantly but it will not go behind my theory of being able to sleep. HOWEVER, I'm removing my personal sleep issues at this point just to be able to potentially get into the market. I will continue to have dual servers for personal usage as much as possible and maybe in the future - IF I CAN PROVE QUALITY / TRACK RECORD, I will revisit this sort of higher end setup. Again, I based this on your most valued feedback and I appreciate you all for being 'gentle' - it was easy to consume and easy to accept.
I was hoping to bring something out for black friday (obviously it's going ot be beta and in small quantity) but I still feel like my prices would be higher. These are my proposed prices if you are willing to share your feedback or questions. Instead of using OVH, I will be using Hetzner with this new plan and I'll probably just start with one server to see how the market goes.
I have a dilemma on if I should offer 500GB packages or 1TB packages. For 500GB packages, would $4 be too much per month? This is per month billing only since I need to make sure this works before I commit to yearly deals (not going to happen for some time). Again it's not that I cannot fund but obviously I would not like to make losses for a long period of time.
I realize a lot of lowend users may want rsync capabilities (is this true that most of you think rsync is extremely important?) - I'm more concerned about security and opening up SSH too much for that. SFTP was an easy configuration which didn't involve allowing uses to run shell commands.
So the question is, why would anyone want to purchase these plans from me?
These are my potential answers.
Controlled growth, controlled support - I'm not mass selling because I want to provide a service that obviously is profitable but I want to try to be there for my customers. I'm doing server to server at a time. No massive expansion and then ghosting people.
Instead of the original OVH bandwidth of 250 mbit/s shared. Its going to be the standard 1 gbit/shared connection.
Price point is now at $4 for 500 GB compared to $5 for 100 GB. What hasn't changed is that the storage is dedicated - no overselling there - what's yours is yours - fill it with 0s if you have to! It's also RAID 1. The difference - no backup in the event of a catastrophic 2 disk failure so no one gets the 'get good sleep at night' dual server setup.
Guaranteed config is SFTP access and HTTPS for certain aspects (more in point #5). I hope to test rsync (based on your feedback on if not having rsync is a deal breaker). I'm worried that rsync would also be hard on the IO if this is a shared server that has saturated users (not going to happen in the start) - the servers would have enterprise disks but I wouldn't want a crawling server if someone keeps hitting rsyncs every 30 seconds - because that just wouldn't be good for all customers.
Valued added service innovations - it's not going to be just SFTP. I cannot give out all the secrets yet but sections of your folder structure is split into private and public. If you place 'stuff' in your public folder, it's accessible via HTTPS. I have a very basic image gallery view for example. You also have a password protected 'video' section - not as advanced as plex or jellyfin but at least you wouldn't need a separate VPS to have your own personal private collection (password protected of course, and mostly streamable via the web once your video is browser playable). This is a work in progress - meaning there's more to be done as the feature set builds but the basics are up and running. I do have a transcoding idea I'd like to work on as well (not available but hopefully eventually since I doubt at this point I expect to max out the CPU).
P.S Based on your feedback, I may offer this before BF simply because I doubt I can compete on BF with the insane deals that other providers can do.
@risharde said:
I have a dilemma on if I should offer 500GB packages or 1TB packages. For 500GB packages, would $4 be too much per month?
Personally, I'd say so as for instance interserver is $6 per TB normally, but anyone who got the birthday $3 per TB deal can grow at the same price (I think).
I realize a lot of lowend users may want rsync capabilities (is this true that most of you think rsync is extremely important?) - I'm more concerned about security and opening up SSH too much for that. SFTP was an easy configuration which didn't involve allowing uses to run shell commands.
I guess that rules out using it as a borg backup server.
Hi @ralf, thanks for the insight there, really appreciate it. Hmm I know I've seen a lot of threads about borg so I'll try to read up more about it to see what it requires and see if its possible to make it happen. I hope to get around to testing rsync in a day or two (hopefully max). I realize I will have to try to entice customers not just on price (which is a bit higher) so I want to reiterate while it's 'backup' storage, it's going to come with a few bells and whistles which may allow maybe a niche to hop on and not have to worry too much about having an additional VPS to start with (with the 'video' section). But of course, it can be mounted from another VPS.
Does anyone have preference of location between Germany and Finland? I see Germany seems to be located more central to Europe so I'm making a wild assumption that there's a few ms less to reach it
@risharde said: Price point is now at $4 for 500 GB compared to $5 for 100 GB.
This is way too expensive to be interesting. Hetzner Storage Box is a little over $2 per TB in the larger sizes, and it is raid 6. Storage Share is raid 6 and backed up daily and supports nextcloud, and is around $3 per TB.
Comments
You cunt sleep I see.
Why?
Ah-Ha!
This will not help with sleeping.
You should do 21 burpees every morning, for better sleep at night.
Webhosting24 aff best VPS; ServerFactory aff best VDS; Cloudie best ASN; Huel aff best brotein.
I would need to drink a lot of fizzy pop to do 21.
raid0, raid1 and police raid
(depending on the content hosted)
Thanks to everyone who commented in, I took a few days break away from the comments to think more about the idea. I'm convinced the idea will not work as many of you mentioned in terms of it simply being a storage alternative to much larger organizations. I'm still going to continue my second phase which is software/feature related. I have to re-iterate that I'm solving a personal problem which I was hoping to share with an expected smaller subset of the LE community (because I know it's expensive). I'm also very much aware of price sensitivity - I'm a bit surprised that many of you put your trust in the large organizations (I know they have good tech and skills but I just cannot see myself putting my trust in for example Amazon - I know they have come a long way and have become better at what they do but neither can I say I feel comfortable putting my data only on Google as well) NEVERTHELESS I APPRECIATE THE FEEDBACK SINCE IT'S A LEARNING LESSON!
The project basically started off solving my own challenge which is trust only to be shut down by others on the merit of trust as well (ironically) so I'm still motivated to solve the problem I had previously which is data loss. This week the provider I store my backups with had to go down due to node issues and it still seems to me that this project again is necessary for me - maybe one day it will be worth merit and some of you will reconsider my stand point. I've said this throughout the years actually that I have very little confidence in RAID10 which to most of the IT world seems to be heressy but say what - everybody has their own taste. I'm just really trying to sleep good at night (as funny as that sounds and as much holes poked into the implementation) - I've solved this partially for myself.
There's a next part I'm trying to solve for myself which is ease of use from another angle and I'm being very careful not to release that info yet but I'm hoping it will be relatively the exciting part to see what can be done with the storage - I know for sure others will poke holes in that as well but I'm still doing it because I really need it. I'll still likely make offers available - and if people want to come on board - by all means that would be great - even if it's just enough to fill one batch of nodes - it would still be awesome. IF no one use it, bummer, but not the end of the world. I love building things and again the most enjoyable part (other than actually a profit - while small) is actually people using and liking what I build - can't say I've achieved that here as yet but I'll stay optimistic.
Thanks to all the testers so far - you pretty much have storage space indefinitely for free and also thanks to all those who commented in with feedback (even if I saw it as negative - it's still probably realistic so I appreciate it!)
Will keep everyone updated as things hopefully progress!
ONDYNAMIC NETWORKS | Chat & Support
BTW, on OGF, @Neoon mentioned seaweed FS a while ago that sounded interesting and might solve some of the things you're trying to solve.
https://lowendtalk.com/discussion/comment/3437916/#Comment_3437916
Never used that on that cluster, had issues with the mount and people reported suddenly corrupt files.
Free NAT KVM | Free NAT LXC
You sure sound passionate about this. If you consider open sourcing your code down the road I'd love to look at it. I'm sure it'll be interesting.
Going to try it again, in a few days, I let you know.
Time to setup my mega cluster.
https://github.com/Ne00n/seaweed-spawner-3000
Gonna be helpful once more.
Free NAT KVM | Free NAT LXC
I noticed this discussion today only, it was interesting read. @risharde does seems to be passionate and defending his idea was great. But later on accepting that it may not work was kind of smart.
Whatever you are doing it, keep doing just try to think what problem are you trying to solve which is not solved by others already or atleast you can do better than them. Even if it fails, you will still gain a lot in terms of knowledge and experience which you might put to good use elsewhere in future.
Hi everyone! I assure you, I have not been sleeping but I have been a bit quiet as I worked through making this make sense. My primary goal was to build an environment where data was replicated to two physical servers - something that seems to trash the whole idea of being 'cheap' or 'low end'. Again, I hope I considered each of your points correctly and I am going to downgrade my idea back to a single server with RAID. This will half the cost instantly but it will not go behind my theory of being able to sleep. HOWEVER, I'm removing my personal sleep issues at this point just to be able to potentially get into the market. I will continue to have dual servers for personal usage as much as possible and maybe in the future - IF I CAN PROVE QUALITY / TRACK RECORD, I will revisit this sort of higher end setup. Again, I based this on your most valued feedback and I appreciate you all for being 'gentle' - it was easy to consume and easy to accept.
I was hoping to bring something out for black friday (obviously it's going ot be beta and in small quantity) but I still feel like my prices would be higher. These are my proposed prices if you are willing to share your feedback or questions. Instead of using OVH, I will be using Hetzner with this new plan and I'll probably just start with one server to see how the market goes.
I have a dilemma on if I should offer 500GB packages or 1TB packages. For 500GB packages, would $4 be too much per month? This is per month billing only since I need to make sure this works before I commit to yearly deals (not going to happen for some time). Again it's not that I cannot fund but obviously I would not like to make losses for a long period of time.
I realize a lot of lowend users may want rsync capabilities (is this true that most of you think rsync is extremely important?) - I'm more concerned about security and opening up SSH too much for that. SFTP was an easy configuration which didn't involve allowing uses to run shell commands.
So the question is, why would anyone want to purchase these plans from me?
These are my potential answers.
Controlled growth, controlled support - I'm not mass selling because I want to provide a service that obviously is profitable but I want to try to be there for my customers. I'm doing server to server at a time. No massive expansion and then ghosting people.
Instead of the original OVH bandwidth of 250 mbit/s shared. Its going to be the standard 1 gbit/shared connection.
Price point is now at $4 for 500 GB compared to $5 for 100 GB. What hasn't changed is that the storage is dedicated - no overselling there - what's yours is yours - fill it with 0s if you have to! It's also RAID 1. The difference - no backup in the event of a catastrophic 2 disk failure so no one gets the 'get good sleep at night' dual server setup.
Guaranteed config is SFTP access and HTTPS for certain aspects (more in point #5). I hope to test rsync (based on your feedback on if not having rsync is a deal breaker). I'm worried that rsync would also be hard on the IO if this is a shared server that has saturated users (not going to happen in the start) - the servers would have enterprise disks but I wouldn't want a crawling server if someone keeps hitting rsyncs every 30 seconds - because that just wouldn't be good for all customers.
Valued added service innovations - it's not going to be just SFTP. I cannot give out all the secrets yet but sections of your folder structure is split into private and public. If you place 'stuff' in your public folder, it's accessible via HTTPS. I have a very basic image gallery view for example. You also have a password protected 'video' section - not as advanced as plex or jellyfin but at least you wouldn't need a separate VPS to have your own personal private collection (password protected of course, and mostly streamable via the web once your video is browser playable). This is a work in progress - meaning there's more to be done as the feature set builds but the basics are up and running. I do have a transcoding idea I'd like to work on as well (not available but hopefully eventually since I doubt at this point I expect to max out the CPU).
P.S Based on your feedback, I may offer this before BF simply because I doubt I can compete on BF with the insane deals that other providers can do.
Thanks everyone!
ONDYNAMIC NETWORKS | Chat & Support
Personally, I'd say so as for instance interserver is $6 per TB normally, but anyone who got the birthday $3 per TB deal can grow at the same price (I think).
I guess that rules out using it as a borg backup server.
Hi @ralf, thanks for the insight there, really appreciate it. Hmm I know I've seen a lot of threads about borg so I'll try to read up more about it to see what it requires and see if its possible to make it happen. I hope to get around to testing rsync in a day or two (hopefully max). I realize I will have to try to entice customers not just on price (which is a bit higher) so I want to reiterate while it's 'backup' storage, it's going to come with a few bells and whistles which may allow maybe a niche to hop on and not have to worry too much about having an additional VPS to start with (with the 'video' section). But of course, it can be mounted from another VPS.
ONDYNAMIC NETWORKS | Chat & Support
Here's the borg stuff: https://borgbackup.readthedocs.io/en/stable/usage/serve.html
I'd never seen SSH keys used this way, but it's really interesting.
Does anyone have preference of location between Germany and Finland? I see Germany seems to be located more central to Europe so I'm making a wild assumption that there's a few ms less to reach it
ONDYNAMIC NETWORKS | Chat & Support
This is way too expensive to be interesting. Hetzner Storage Box is a little over $2 per TB in the larger sizes, and it is raid 6. Storage Share is raid 6 and backed up daily and supports nextcloud, and is around $3 per TB.