Photography, geek, gadget stuff and life.
An alternative to Drobo. The MediaSonic StarTech HFR2-SU3S2 ProRaid USB 3
Being a photographer, storage is incredibly important; storage that can expand and offer protecting is crucial.
When it comes to direct access storage devices that offer built in disk failure redundancy features and multi drive capabilities the biggest name is Drobo by Data Robotics.
However, I lived with a Drobo V2 for just over a year. This version had the USB2 and Firewire 800 interfaces. In the beginning it was great. Drobo was small, quiet and kept my files safe and protected. But I was soon wanting more from the diminutive device. I needed faster random access, faster sequential data throughput and finally I wanted an transfer interface that wasn’t based on 10 year old technology (USB and FireWire). So it was time to move on and I sold it.
18 months later there was a Drobo shaped space next on my desk. I still wanted all those features which the Drobo couldn’t offer. I looked at the Drobo’s newer siblings; the Drobo S and Drobo FS but reviews slated them for the slow speed and their inadequate ability to maximize their eSata and USB 3 interfaces. Plus they were notoriously expensive.
The search continued for many months when I stumbled on a small device from MediaSonic for around £100 (approx $160 US).
The device has a very silly name, it’s called the HFR2-SU3S2. Even with the silly name it does have a lot features. It has an eSata and USB 3 interface, 4 drive bays and a choice of RAID 0, 1,5 and 10.
This is how I’ve configured it:
- Three 2TB Western Digital drive at 5200RPM
- USB 3 interface
- RAID 5 configuration giving me 4TB of usable storage
Using the USB 3 interface it’s as speedy as any SATA drive in my PC; I was hoping it would be faster since I have it configured in RAID 5. I was expected higher performance since the data is spread across the three 2TB drives I have installed but without doubt the results are faster than the Drobos.
There is no software provided and the instructions are short; however this was not an issue for myself as I have a lot of experience with these devices, but I can imagine it would be a bit intimidating for the absolute beginner.
I have read other comments about the HFR2-SU3S2 being quite noisy; I have found it to be very quiet. It is by no means silent. I would describe the noise level to be on the ‘very quiet’ scale. I set the fans on auto and it just does the rest!
In the box, you get a USB3 and a eSATA cable. Both are quite short so if you plan on having the unit on your desk and your computer is underneath your desk, then you may have to purchase a longer lead like I did. I got a 2m eSATA cable for only £2.00
The lights on the front are bright but not distracting. When the drives are being accessed the activity lights turn purple. There have been comments from other users saying it’s too bright but I disagree. While they are noticeable on my desk they don’t annoy me.
Overall, this is a magnificent little device that has little compromise when compared to the more expensive Drobo. The only limitation I see is all the drives have to be matched (you can mix up manufacturers and sizes on the Drobo).
Buy it, it’s cheap and offers an expandable means of storage with redundancy.
-
#3 written by Myles Fudge 4 months ago
Interesting article, especially since I'm just beginning to set up a ZFS RAIDz storage system using FreeNAS. And… I currently own and use the 8-bay model of the MediaSonic ProRAID box and the 8-bay model of the Synology Diskstation. I have long contemplated the safety of my data, having experienced numerous hard drive failures and personal data loss. I have to second Ben's excellent comments and advice: RAID IS NOT A BACKUP. Being a current IT guy and amateur terrestrial and astro-photographer (read: data-collector), these are my thoughts about storage reliability.
It's old news now, that RAID5 has become quite a dangerous storage solution when using very large hard drives. That is the main reason why I moved from MediaSonic RAID5 to Synology RAID6, and now I am moving over to ZFS for even more data reliability. My MediaSonic RAID5 was 4x2TB drives. My Synology SHR RAID grew from 4x3TB+2x2TB, to 8 drives with 3+3+3+3+2+2+4+4 TB (24 TB total, or 18TB formatted with 1-drive redundancy). I will be rebuilding it shortly with 2-drive parity (RAID6).
My primary concerns are RAID5 rebuild time for an array with 2TB – 4TB hard drives (24+ hours), and Time Limited Error Recovery/deep recovery cycle of hard drives. If you build your RAID5 setup with many smaller hard drives such as 250GB or 320GB drives then the rebuild time is much quicker than it would otherwise be with very large terrabyte hard drives. While using RAID5 with 2,3,4 TB hard drives you risk disaster in that while your RAID is rebuilding from a single hard drive failure (or deep recovery cycle), an additional hard drive failure (or additional deep recovery cycle) during that long rebuild/recovery period will wipe out your entire RAID. It is much wiser to move to double parity RAID6 (or better) so you can withstand two (or more) simultaneous hard drive failures and still not lose your data.
Regarding hard drive deep recovery cycles, if during the normal operation of your RAID5 array one of your hard drives detects an error, the drive will automatically become unavailable for up to two minutes while the hard drive firmware tries to recover the data. Your RAID box will then "drop" that unavailable drive from your array usually between 7-15 seconds. If this simple read error occurs (while your RAID5 is stressed during the rebuild stage of a drive offline/replacement) then you will have lost 2 drives and your entire array will be lost. To avoid the perfectly-normal deep recovery cycle, you must purchase much more expensive Enterprise-grade hard drives which utilize Time Limited Error Recovery and will not spend up to two minutes trying to recover data. [Hopefully hard drive companies will enable this simple TLER feature (perhaps by simply setting a jumper) for all classes of drives instead of forcing us to purchase Enterprise-class drives].
I have not used a Drobo box, but I am aware of their considerable cost ($2,700 vs $1,050 for Synology DS1812+). However the MediaSonic ProRAID 8-bay box was only $350 but does not support RAID6. For real data safety, I would finish with this:
"An alternative to Drobo. Vastly superior ZFS using FreeNAS and less-expensive off-the-shelf hardware."
-
#4 written by Myles Fudge 4 months ago
I'm not sure if the FaceBook Social Plugin made this reply post only to Facebook, or also as a reply to this Tech website posting…
Interesting article, especially since I'm just beginning to set up a ZFS RAIDz storage system using FreeNAS. And… I currently own and use the 8-bay model of the MediaSonic ProRAID box and the 8-bay model of the Synology Diskstation. I have long contemplated the safety of my data, having experienced numerous hard drive failures and personal data loss. I have to second Ben's excellent comments and advice: RAID IS NOT A BACKUP. Being a current IT guy and amateur terrestrial and astro-photographer (read: data-collector), these are my thoughts about storage reliability.
It's old news now, that RAID5 has become quite a dangerous storage solution when using very large hard drives. That is the main reason why I moved from MediaSonic RAID5 to Synology RAID6, and now I am moving over to ZFS for even more data reliability. My MediaSonic RAID5 was 4x2TB drives. My Synology SHR RAID grew from 4x3TB+2x2TB, to 8 drives with 3+3+3+3+2+2+4+4 TB (24 TB total, or 18TB formatted with 1-drive redundancy). I will be rebuilding it shortly with 2-drive parity (RAID6).
My primary concerns are RAID5 rebuild time for an array with 2TB – 4TB hard drives (24+ hours), and Time Limited Error Recovery/deep recovery cycle of hard drives. If you build your RAID5 setup with many smaller hard drives such as 250GB or 320GB drives then the rebuild time is much quicker than it would otherwise be with very large terrabyte hard drives. While using RAID5 with 2,3,4 TB hard drives you risk disaster in that while your RAID is rebuilding from a single hard drive failure (or deep recovery cycle), an additional hard drive failure (or additional deep recovery cycle) during that long rebuild/recovery period will wipe out your entire RAID. It is much wiser to move to double parity RAID6 (or better) so you can withstand two (or more) simultaneous hard drive failures and still not lose your data.
Regarding hard drive deep recovery cycles, if during the normal operation of your RAID5 array one of your hard drives detects an error, the drive will automatically become unavailable for up to two minutes while the hard drive firmware tries to recover the data. Your RAID box will then "drop" that unavailable drive from your array usually between 7-15 seconds. If this simple read error occurs (while your RAID5 is stressed during the rebuild stage of a drive offline/replacement) then you will have lost 2 drives and your entire array will be lost. To avoid the perfectly-normal deep recovery cycle, you must purchase much more expensive Enterprise-grade hard drives which utilize Time Limited Error Recovery and will not spend up to two minutes trying to recover data. [Hopefully hard drive companies will enable this simple TLER feature (perhaps by simply setting a jumper) for all classes of drives instead of forcing us to purchase Enterprise-class drives].
I have not used a Drobo box, but I am aware of their considerable cost ($2,700 vs $1,050 for Synology DS1812+). However the MediaSonic ProRAID 8-bay box was only $350 but does not support RAID6. For real data safety, I would finish with this:
"An alternative to Drobo. Vastly superior ZFS using FreeNAS and less-expensive off-the-shelf hardware."
-
- Comment Feed for this Post
I was reading through your posts and wanted to comment and point something out that I see people do far to often, and that is to consider a RAID as a backup. Its not. RAID is for speed and HA (high availability). It should never be used as a primary backup solution (especially RAID5). The problem is bit error rates and the controller and PCs ability to handle them. Without ECC ram and a checksumming filesystem like ZFS you can write bad data and never know it until you check the file and find it corrupted or your array drops. Raid5 rebuilds put the whole array at risk because a single bad block in the parity information will kill the rebuilt and drop the array. Then "poof" goes the data.
As an ex-IT guy and fulltime photographer I am constantly trying to improve my backup solutions (well, photographers helper, lol, its the wife who's the "real" boss).
Currently I use a LTO3 tape drive (400GB/800GB tapes fit right around 400GB of RAWs because they do not compress well). I also back up to bluray and JPGs to smugmug and keep the files alive on my drive arrays until space is needed, then they get wiped (but still exist on the LTO tapes, Blurays and online on smugmug).
You can find the LTO drives for a few hundred dollars now if you watch craigslist or ebay closely and a bluray drive is 60-90$ with disks running 1-2$ each.
Using those two methods combined costs about.08 cents a GB or in my case about 3.75$ per wedding. If you do say 25 weddings a year you would be at 93$ a year for double backups on each event even if you wiped them off the HDDs right after the backup.
I love arrays for speed (raid0 with file or block level nightly syncs to a backup drive, *if* your willing to lose a days worth of data) or RAID10, or RAID6. I guess RAID5 could be ok if you also treat it like RAID0 and sync it to another drive nightly. Or just use ZFS RAIDz.