How to do NAS?

[Kilrah]

Latest ransomware encrypting networkshares have proven how bad a NAS is as a backup strategy.

Well nothing says that the backup NAS has to be be constantly online.

[Kalvin]

Latest ransomware encrypting networkshares have proven how bad a NAS is as a backup strategy.

Well nothing says that the backup NAS has to be be constantly online.

Or, nothing says that the backup NAS has to be accessable by the users. The backup NAS could operate without user access, using a file system with snapshot capabilities.

[Kilrah]

The snapshot capabilities go against the "limited cost" option though, since you need significantly more storage space than the actual data when doing that i.e. more/bigger drives.

[amspire]

Kalvin, when the Nas has read-only snapshots and no network drives?

[amspire]

The snapshot capabilities go against the "limited cost" option though, since you need significantly more storage space than the actual data when doing that i.e. more/bigger drives.

You need more space, but it is usually not that much extra space.

[Kalvin]

Kalvin, when the Nas has read-only snapshots and no network drives?

Of course there are network drive(s), otherwise it wouldn't be NAS. The NAS can have a file system such as ZFS providing snapshots and ability to restore accidentally deleted files. The NAS will be mirrored so that the user cannot access those backup devices directly, instead the backup devices will just incrementally mirror the NAS's contents as backup purposes only. Should the NAS have a mechanical or electrical failure, the data is still available on the backup devices. The backup devices will be used to rebuild the NAS again. During the rebuild phase the backup devices can be mounted read-only, but the data is still available to the user. When the data has been restored from the backup into the repaired/replaced NAS, the system is fully operational. Typically a NAS has a RAID system built in using multiple hard disks, but instead of multiple hard disks I would like to use multiple, physically distributed backup devices providing the redundancy and recovery ability should the NAS experience any failure.

[amspire]

If I am using a nas as a backup, I don't have any network drives. If you are using a nas  as an on line working storage, it is not the backup. I use rsync to sync files to the backup and and winscp with no stored password to access files on the bad boxes. I initiate sync from the offsite nas to the onsite nas so that is impossible for a virus that has total control of the network to touch the offsite nas.

I have had experience with cryptoviruses attacking a nas with a network drive and it could not touch the snapshots or the offline backup. But following this experience, I have disabled all network shares anyway. I do not allow users to browse the backup nas boxes.

It is all about strategy. As I have said, traditional rotational backup is worse unless you realise immediately that files have been encrypted, corrupted or deleted. This is often not the case. With rotational  daily  + monthly backups, if you don't detect the damage (virus, accidental or deliberate)  within 7 days, you can loose files from the current month. A good nas backup with zfs or btrfs snapshots does not have this problem at all.

Anyone who thinks a rotational backup is a total backup is as deluded. No single method is perfect.

You have suggested a different architecture. Fine, it is all strategy. Just stop suggesting that there is only one way to use a nas.

[edavid]

I recently had to figure out this exact problem.
We previously had a WD Mybook Live NAS, which was a complete piece of shit. The transfer speeds were downright awful, you always had to wait for the disk to spin up when you tried to access the files, terrible UI and on top of that, it bricked itself when I tried updating the firmware.
I ended up pulling the HDD out of the enclosure and had to jump through a bunch of hoops to recover the files from the disk (many thanks to the good people on #eevblog for helping me with that).

Funny, I am completely satisified with my WD MB Live Duo:

1. Transfer speeds are around 30MB/s, good enough for my use (it's a low power, fanless device)

2. Disk spindown time can be configured, or it can be turned off

3. UI seems OK to me, but I never use it, so who cares

4. No trouble with firmware updates, but I doubt I will apply any more of them, since I can just use apt-get for any updates I actually want

However, it obviously wouldn't meet the OP's needs, so why even mention it 

[legacy]

LTO Ultrium tape backup ?
during the weekend ?

[Kalvin]

You have suggested a different architecture. Fine, it is all strategy. Just stop suggesting that there is only one way to use a nas.

Maybe this is a language issue as English is not my native language, but my intention was not suggest anything per se, only present an idea what I have had in my mind and wanting feedback and criticism from other members about pros and cons.

The idea I am thinking of is similar to what Google is doing ie. using cheap hardware and distributed redundancy:
http://www.techworld.com/storage/googles-storage-strategy-467/

The hard drives
Given this why not have a large disk server used by the clustered Linux machines? It's cost and reliability that drives this. A disk server is expensive and, as a single box, is vulnerable. Getting the hard drives with the PC servers means that the data is stored across hundreds if not thousands of drives. Google replicates data three times for redundancy. It can afford to be cavalier about hardware failures. So a drive fails. Log it, switch queries on that data to a replica and move on. It's all pretty instant.

There isn't even RAID protection. In a way the Google cluster architecture is similar to the RAIN storage idea, a redundant array of inexpensive nodes. (Techworld mentioned RAIN here. Exagrid is a supplier with RAIN storage product ideas which Techworld discussed recently here.)

The drives are IDE drives and not SCSI, which would be more expensive. Google spends more time reading files than waiting for them to be read. Latency is not that great an issue so having lightning fast 15,000pm SCSI drives is not a requirement. In 2001, 5400rpm 80GB maxtor IDE drives were mentioned as being used by Google
...
Google gets its system reliability from software and hardware duplication. It uses commodity PCs to build a high-end computing cluster.

[amspire]

Kalvin, when the Nas has read-only snapshots and no network drives?

Of course there are network drive(s), otherwise it wouldn't be NAS.

NAS is network attached storage, but you can choose how you want to attach. That is why NAS boxes usually have a number of available protocols including SMB (Windows shares) and NFS that you can enable or disable. I use rsync and ssh and I have SMB and NFS disabled for my backup NAS boxes.

I do this because I know the cryptoviruses definitely do scan the local network for Windows shares, and there is no need to have Windows shares enabled on a backup NAS. At the moment, they do not try rsync connections. They could but right now they don't.

There is no law that says you must have windows shares enabled on a NAS box.

[legacy]

and no tape backup? anyone?

[legacy]

well, LTO2 costs 150-180 euro brand new
it's wide-SCSI band, it means 40Mbyte/sec
and a box of five cartridges costs 50 euro(1)
10 euro per cartridge and you have 200Gbyte
at least per cartridge (400Gbyte compressed)

(1) IBM original cartridges

[legacy]

Factoring inflation and tape drive failure

very; thoroughly, both false

[legacy]

I can buy a 4TB 2.5'' HDD for $130, equivalent to $0.033 per GB

and a tape lasts longer (and with less troubles)
than an electromechanical hard drive

[madires]

Tape backup for SOHO applications doesn't make much sense nowadays. One thing people forget is the increasing amount of data to backup. When you do a few full backups per year as recommended, you'd become a tape jockey. The better way is to buy current hard disks plus decent USB-HDD boxes. When they become too old or small, simply get new ones.

[legacy]

it depends on your needs

here I am happy with my LTO2, DLT and DAT
they are good when you have to send/receive
data to/from abroad through a paper postage
(super cheap price through UPS/Fedex)

can you do it with mechanical hard drives ?

I am also experimenting CompactFlash cartridges
through FireWire 400/IEEE 1394a interface (by Lexar)

the bandwidth is not good, 20Mbyte/sec
but I can stack two units in parallel

a better option should be ATA Flash CompactFlash
and PCMCIA Cards Internal interface made by Delock

and then interfacing it to a pATA to sATA
ora pATA to Ultra Wide SCSI interface

the bandwidth will be ~ 50 Mbyte/sec

the cost, for a qty=10 of 64Gbyte each
it's 25 euro per CF-cartridge (by Kingston)
something like 250 euro per year

they are super marvelous for my own monthly repository
and I can ship a CF-cartridge to colleagues in case of need
(or get a backup back from them)

comfortable and nice, also supported by my NAS

[legacy]

my last purchase confirms you are right about prices:
I got eight hard drives, 40 euro each, for a qty=8 sATA2
Barracuda 7.2K rpm, with 64Mbyte of data buffer, 1Tbyte

[SeanB]

and no tape backup? anyone?

A modern LTO5/6 tape drive costs 20x more than consumer HDDs. Even if the tapes are free (which is not), to recover the cost of the tape drive, I need to run it for 40 years assuming if otherwise I retire backup hard drives every 2 years.
Factoring inflation and tape drive failure, I do not think I can ever recover the additional cost at all. The only time where economically a tape drive makes sense is if you have tons of data that the lower cost per TB for tapes actually offsets the initial cost of a tape drive, or you definitely need to store data for decades as if you are running a bank or a digital library.

Do not forget the tapes have a limited life, and you will need a cleaning tape as well for the drive. Interesting RFID tag they put in there though, and a really intricate tape mechanism as well, with all those pivot points and the integrated spool brake as well.

[ericloewe]

I'll take this opportunity to plug FreeNAS.

Let's start by addressing some issues raised here:

• Ransomware - With ZFS snapshots, you're laughing in the face of the moron who thought he could hold your data hostage. It's trivial to rollback to an earlier state and FreeNAS even provides integration with Windows' Previous Versions feature, for easy access to specific files in old snapshots
• Data integrity - ZFS is pretty much proven in this regard, with much more real-life experience than btrfs. ZFS goes to immense lengths to providew reliable storage (If it's worth storing, it's worth storing properly)
• Transfer speeds - ZFS may sound like a resource hog, but it's anything but slow outside of pathological workloads. Real-life performance for file-sharing workloads is in the multi-gigabit region with slow, low-TDP processors. Fast systems can easily saturate 10GbE

Now, FreeNAS and ZFS are not forgiving of hacked-together systems with little though put into them. Additionally, the community strongly pushes proper hardware choices that are uncommon among enthusiasts - no gaming rigs over there, the servers are really servers.

A common misconception is that cost for FreeNAS recommended hardware is much higher - this isn't true at all. Cost is higher than for crap consumer stuff that's held together by a prayer and bloated Windows-only drivers, but it's on par with good consumer desktop components.

Some research is required and it involves work input, but the results are much better than what you can expect from typical NAS boxes. The community has made an effort to compile all the required information, and most of the important information is now centralized in the forum's Resources section. Members there will happily nudge new users in the right direction, but hand holding is not provided.

Full disclosure: I'm a mod at the FreeNAS forums.

Edit: Useful links:

https://forums.freenas.org

Resources section of the forum:
https://forums.freenas.org/index.php?resources/

[Scrts]

I'll take this opportunity to plug FreeNAS.

Let's start by addressing some issues raised here:

• Ransomware - With ZFS snapshots, you're laughing in the face of the moron who thought he could hold your data hostage. It's trivial to rollback to an earlier state and FreeNAS even provides integration with Windows' Previous Versions feature, for easy access to specific files in old snapshots
• Data integrity - ZFS is pretty much proven in this regard, with much more real-life experience than btrfs. ZFS goes to immense lengths to providew reliable storage (If it's worth storing, it's worth storing properly)
• Transfer speeds - ZFS may sound like a resource hog, but it's anything but slow outside of pathological workloads. Real-life performance for file-sharing workloads is in the multi-gigabit region with slow, low-TDP processors. Fast systems can easily saturate 10GbE

Now, FreeNAS and ZFS are not forgiving of hacked-together systems with little though put into them. Additionally, the community strongly pushes proper hardware choices that are uncommon among enthusiasts - no gaming rigs over there, the servers are really servers.

A common misconception is that cost for FreeNAS recommended hardware is much higher - this isn't true at all. Cost is higher than for crap consumer stuff that's held together by a prayer and bloated Windows-only drivers, but it's on par with good consumer desktop components.

Some research is required and it involves work input, but the results are much better than what you can expect from typical NAS boxes. The community has made an effort to compile all the required information, and most of the important information is now centralized in the forum's Resources section. Members there will happily nudge new users in the right direction, but hand holding is not provided.

Full disclosure: I'm a mod at the FreeNAS forums.

Edit: Useful links:

https://forums.freenas.org

Resources section of the forum:
https://forums.freenas.org/index.php?resources/

I was always thinking, that the power consumption of the PC server is significantly higher than a dedicated NAS. Will check that forum if somebody made any comparisons.

[technix]

I'll take this opportunity to plug FreeNAS.

Let's start by addressing some issues raised here:

• Ransomware - With ZFS snapshots, you're laughing in the face of the moron who thought he could hold your data hostage. It's trivial to rollback to an earlier state and FreeNAS even provides integration with Windows' Previous Versions feature, for easy access to specific files in old snapshots
• Data integrity - ZFS is pretty much proven in this regard, with much more real-life experience than btrfs. ZFS goes to immense lengths to providew reliable storage (If it's worth storing, it's worth storing properly)
• Transfer speeds - ZFS may sound like a resource hog, but it's anything but slow outside of pathological workloads. Real-life performance for file-sharing workloads is in the multi-gigabit region with slow, low-TDP processors. Fast systems can easily saturate 10GbE

Now, FreeNAS and ZFS are not forgiving of hacked-together systems with little though put into them. Additionally, the community strongly pushes proper hardware choices that are uncommon among enthusiasts - no gaming rigs over there, the servers are really servers.

A common misconception is that cost for FreeNAS recommended hardware is much higher - this isn't true at all. Cost is higher than for crap consumer stuff that's held together by a prayer and bloated Windows-only drivers, but it's on par with good consumer desktop components.

Some research is required and it involves work input, but the results are much better than what you can expect from typical NAS boxes. The community has made an effort to compile all the required information, and most of the important information is now centralized in the forum's Resources section. Members there will happily nudge new users in the right direction, but hand holding is not provided.

Full disclosure: I'm a mod at the FreeNAS forums.

Edit: Useful links:

https://forums.freenas.org

Resources section of the forum:
https://forums.freenas.org/index.php?resources/

I was always thinking, that the power consumption of the PC server is significantly higher than a dedicated NAS. Will check that forum if somebody made any comparisons.

PC server does have much greater flexibility even though it is being used as a NAS. I have a NAS/router combo server that have all those features:

• Quad-core Intel Core 2 Quad Q9300, 8GB DDR2-800 ECC memory
• 9TB of hot-swappable storage with RAID-6 hardware redundancy (tolerates the failure of 2 drives)
• 6 1000BASE-T with 802.3ad LACP link aggregation, 802.1Q Ethernet VLAN, PPPoE and PPPoEoV
• IPv4 and IPv6 dual stack gateway with IPv6 tunnelling
• Dynamic DNS
• DNSCrypt with cache for local network
• Can be accessed through SMB/CIFS (Windows), AFP (macOS, supports Time Machine backups), NFS and rsync
• KVM virtual machines for hosting long tasks I don't want to mind on my own workstation

[technix]

Quad-core Intel Core 2 Quad Q9300, 8GB DDR2-800 ECC memory

How did you get ECC to work with consumer Core 2 Quad? Are you using a 771 mobo?

I used the Asus P5BV-C. It is a LGA775 one with Intel 3200 MCH intended for those Xeon 3300 series, but since it uses the same socket as consumer Core 2 Quad, and back then the memory controller is in the north bridge instead of the CPU itself, I can get a consumer CPU on this server motherboard and still get working ECC memory.

[ericloewe]

I'll take this opportunity to plug FreeNAS.

Let's start by addressing some issues raised here:

• Ransomware - With ZFS snapshots, you're laughing in the face of the moron who thought he could hold your data hostage. It's trivial to rollback to an earlier state and FreeNAS even provides integration with Windows' Previous Versions feature, for easy access to specific files in old snapshots
• Data integrity - ZFS is pretty much proven in this regard, with much more real-life experience than btrfs. ZFS goes to immense lengths to providew reliable storage (If it's worth storing, it's worth storing properly)
• Transfer speeds - ZFS may sound like a resource hog, but it's anything but slow outside of pathological workloads. Real-life performance for file-sharing workloads is in the multi-gigabit region with slow, low-TDP processors. Fast systems can easily saturate 10GbE

Now, FreeNAS and ZFS are not forgiving of hacked-together systems with little though put into them. Additionally, the community strongly pushes proper hardware choices that are uncommon among enthusiasts - no gaming rigs over there, the servers are really servers.

A common misconception is that cost for FreeNAS recommended hardware is much higher - this isn't true at all. Cost is higher than for crap consumer stuff that's held together by a prayer and bloated Windows-only drivers, but it's on par with good consumer desktop components.

Some research is required and it involves work input, but the results are much better than what you can expect from typical NAS boxes. The community has made an effort to compile all the required information, and most of the important information is now centralized in the forum's Resources section. Members there will happily nudge new users in the right direction, but hand holding is not provided.

Full disclosure: I'm a mod at the FreeNAS forums.

Edit: Useful links:

https://forums.freenas.org

Resources section of the forum:
https://forums.freenas.org/index.php?resources/

I was always thinking, that the power consumption of the PC server is significantly higher than a dedicated NAS. Will check that forum if somebody made any comparisons.

Not really. Crap low-end NAS boxes will use less power, but the mid-range stuff is using Intel CPUs these days, so it's going to be very similar to full-fledged servers. Both of my servers (Haswell or Skylake i3, 16GB RAM, few fans, 6 WD Red hard drives) idle below 50W (each, obviously), measured by the UPS). Note that this is with the drives spinning - general consensus is that the longevity hit from spinning down the drives (or even just parking the heads aggressively, like WD Greens / New Blues do) eliminates any power savings.

You could optimize this down with some tradeoffs: Cut the IPMI (remote system management and remote console) and shave a few watts, go with a PSU that's closer to your requirements, with less room for growth, to try and optimize for efficiency...

[Brumby]

Just looking at a second hand QNAP TS-859 Pro....

From what I can gather, it has:
1.66Ghz dual core Atom CPU
1GB DDR2 RAM


Thoughts anyone?