I have a few questions about transition from my Raspberry 3B+ to Synology DS218+.
I am SNO since October 2019. I use Raspberry with 2TB usb hard drive.
I am running everything at my home and have only single one IP.
I am going to buy Synology DS218+ and 2x 12TB WD ultrastar drives.
I will set it up with few UPSes to keep running even if my power goes down.
My Raspberry node is full.
I have 100Mbit/100Mbit internet via optic fiber.
Should I start-up new node on the NAS and wait months to get more reputation and get more data flow and keep my Raspberry running?
Should I rather copy everything from Raspberry to my NAS and continue using my old node ID with good reputation and 2TB of data?
Should I set-up two new nodes for each drive in the NAS? If yes, should I use VM?
Is it a good idea to run multiple nodes under one IP? I heared that I will gain reputation slower and possible data flow will be split between my nodes.
Should I set-up some kind of VPN with different IP for every node?
What’s the best approach to my problem?
The cost of the system as well as the UPS components will never pay for themselves running a Storj node. If you were planning on running the system for other purposes, then there’s no issue… of course, you can purchase what you like… but, purchasing and running the above system is unlikely to break even over the course of several years.
So, I would recommend continuing to run the RPi node… or for more reliability attach the RPi drive to a USB port on the Synology.
I want the NAS for other pusposes as well, so I would buy that even if I would not be able to run storj on it, but I would buy only 2x 2TB drives. So I am thinking about buying more storage with it.
I did some calculations, maybe those are not correct.
Let’s say that every 1TB of hard drive space costs 40USD.
Every TB filled from StorJ is 1.5USD per month. Multiply it by 2 years and you get 36USD which is almost the price of the hard drive itself.
So if the hard drive would get full in half a year and stay online for two more years, it would pay for itself in 2.5 years just with filled space - not counting egress. Even if I would have to replace my drives every 2-3 years I would profit on the egress.
Are my assumptions correct? Maybe I understand it wrong. Just I have Raspberry that payed for itself 2 times in past half a year. So I am just thinking if it would be possible to pay for itself in few years on a bigger scale of 20TB.
Yes. And I like the reasonableness of your calculations.
The WD drives you listed are spec-ed properly for 24/7 operation with 10^15 non-recoverable bit read…
My personal opinion on the purchase of a dedicated NAS
A 2 bay NAS isn’t that useful unless there are external eSATA ports. There are only two RAID options in two drive configuration: RAID 0 and RAID 1. … RAID 0 provides no protection against drive failure and RAID 1 uses 50% of the total drive space for drive failure protection.
One could partition the drives first and then run RAID 5 using the multiple partitions. But that doesn’t make any sense, because if the drive fails, then you still lose your data because you can’t replace a partition of a single drive.
So, I personally wouldn’t buy a dedicated 2 bay NAS product, as it doesn’t make sense to me. However, many people love the specific NAS you’ve listed. So, maybe I’m missing out on something…
I prefer to build my own computing platforms via purchase of individual components. An inexpensive tower case can be re-used over and over again… motherboards replaced easily upon failure or whenever you feel like trying out a new set of features… and plenty of drive bays for expansive addition of 7 or 10 drives as time and funds become available.
So… If I were to buy a NAS, I would look at buying the DS1019+ which at least has the ability to handle 5 drives… allowing one to give ZFS or RAID 10 a try. However, I usually find dedicated solutions to be too restrictive for my tinkerer’s mind.
I assume your 2TB node is full. You’d be able to immediately make more money on that if you give that node more room, so it would be a good idea to copy it to one of the internal disks and have a good reputation node able to receive the full amount of data right away. I would also start the new node on the other HDD right away, even if you limit the size for now, it’s good to have it up and running to get vetted and go through the months of held back amount. You could even choose to start a new node on the RPi 2TB HDD if you have no other purpose for it.
No, Synology has support for docker and you can simply run it in a container. One node for each HDD.
It is if you want to use multiple HDD’s like you do. They will indeed be sharing traffic. I suggest not starting them at the same time. Once the first node is fully vetted, you can start another one. You also want to have at least one node with free space that is vetted so you can receive the full amount of data.
No you shouldn’t, this would introduce latency which will impact your nodes performance and additionally it comes with a monthly cost that will suppress your ability to make money on this.
Egress is probably going to be the bigger part of your income. Currently at about 10% of data stored. So you could count with 1.5+2=3.5USD per month.
Synology doesn’t support ZFS. I would suggest using SHR2 (based on RAID6) on Synology devices, that will allow you to mix drive sizes with dual disk redundancy. But… this is mostly because the other stuff you want to use it for probably deserves some redundancy. Unfortunately, SHR1 (based on RAID5) is not reliable enough anymore.
So while I tend to recommend NAS systems with more bays as well, it all depends on whether you are planning on going with SHR2 or RAID6. If your plan is to use separate HDD’s you might as well get the 2 bay. You can always expand the functionality of that one by buying a DX517 expansion unit. You would pay about the same amount total as you would for the DS1019+ and have 7 bays total instead of 5. There would be a bit of a bottleneck between the main unit and the expansion unit though, but probably not too much of one to be a problem for Storj.
I would setup the NAS so it would see it as 1 big drive of 24TB. This can be done easily as an LVM or as a raid 0. You then only have to run 1 node on your NAS and it will fill up both of the disks.
A setup like this means that if a drives fails you will lose your node (and partly your other data). So in that sense it is a bit riskier.
And as already has been said above. Copy your node to your nas. You won’t have to go through vetting again and you’ll see a quicker ROI on your disks. You can fire up a second node on your pi again afterwards
The databases can be moved to the NAS, the node software can be run from the NAS, the node ID can be copied from the NAS… but I would leave the data on the current drive.
The NAS has USB 3.0 ports. So, if the current drive is hooked to a RPi via USB, it can simply be plugged into the NAS USB port instead.
Just make sure the mount point is accurately changed when starting the node… and then no data needs to be transferred and the NAS itself can be used to run a second node building reputation and escrow before the old node’s drive fails…
Copying the current node data to the NAS just lowers the overall data store.
First of all, thank you all for your suggestions. You helped me a lot deciding what is the best approach.
I was thinking about it but I want to have my personal data backed up - 500GB
Have there movies and tv shows - 4TB for possible plex server or some other kind of video streaming. Which I don’t need to have backed up.
And then the rest will be for StorJ - raids for StorJ are not worth it which was discussed in this forum many times.
Since there is no raid for only part of the disk, I am planning to setup backuping my personal data - 500GB from one drive to another.
Having more bays in my nas would be great, but the price of 4 bay or 5 bay nas with Intel CPU is almost a double than the price of 2 bay DS218+.
I totally agree with you. Making my own solution from parts would be ideal.
I think that it would end up maybe even 2 times cheaper. But I worry about few stuff.
Motherboards, PSU and other parts that you can easily use to build it are not rated for 24/7 use and I worry that it would stopped working after some time.
Also the power consumption would be also a problem. If I can save 200 of dollars on this device, but it would need 50-100W to operate compared to 17W when under load for the DS218+.
If it would be 50W difference in a period of 2 years. It would cost me those 200 dollars or more based on my calculations and price per kWh in my area.
So power consumption is good to include there as well.
At this moment I have only one node. It has a good reputation after more than a half year of operating with minimal downtimes.
I don’t trust that 2TB USB hard drive. Yes it served me well, but I don’t want to rely on it with my only one node ID with good reputation.
So I will probably copy node ID and node data to one of the “datacenter” grade disks (at least that’s how WD calls them) and continue using my well known reliable node on the NAS.
Create a new identity and start a new node on the other drive and maybe keep that USB hard drive pluged in and use it as my 3rd node after the 2nd one gets more reputation and starts gaining more data.
I’m not counting with egress, because it may not be as good as it is now because there might be more people or companies just wanting to back up their data and only sometimes they need to download the backed up data back because of their drives failure.
Maybe it will be the other way around, but noone can know at the moment.
If I want to look at it as an investment, I think I can safely assume that my drives will get full in half a year or one year, but I can’t safely assume that the egress stays at 10%. Of course there will be at least some egress. Which would make my hard drives pay for themselfs even quicker.
If you copy all the data from the node to the new HDD and change the startup command from 2 TB -> 12 TB this node the old node will be able to use all the diskspace available on the NAS.
Therefore the ROI will be shorter, since a new node needs to go through vetting first and you have all the held amounts again.
So I would copy everything over to ulitilize the potential of the new NAS straight away. After everything runs okay you can start a new node on the NAS for the other disk. After that one has gone through vetting you can still start another node on the Pi
… the old hard drive will need to perform 2TB of reads across all platters which will greatly reduce its remaining lifetime. There’s also a non-zero, and quite possibly very large, chance of there being an issue with the data transfer resulting in loss of data or misdirected data followed soon after by DQ of the node ID.
However, it seems likely that we are discussing something that is sometimes described as “6 of one, half a dozen of another” … meaning, the overall difference between both scenarios is likely small.