QNAP TS-120 and TS-212P NAS review

OK, I haven't been blogging for a while. I've been just busy. However, today, I am going to resume and post about two QNAP NASes I've been using for the past 2 months. The 2 Bay QNAP TS-212P (pictured above left) and the single bay TS-120. I'm not going to go into a full review. I'm just going to jot my thoughts on using these two devices.

Now for a bit of a background. As many of my old readers know, I am a fan of home-grown systems; using FreeNAS and re-purposing Pogo Plug devices. Well, for the past year, my FreeNAS box has been at the office. At home, I'm a bit more energy conscious. I've been using various spare Linux boxes to act as SAMBA and TimeMachine shares. I've also used a few Pogo Plug in stock and ArchLinux form. However, my TimeMachine (TM) backups have been a bit flakey. Every other two or three weeks, I would get a corrupted TM image and would have to re-start. Furthermore, I had two pogoplugs fail on me. So when the QNAP went on sale during Christmas, I figure I'd give it a try.

I got the TS-120 for around $80 and the 2 bay TS-212p for $110. They are similar yet different. Both have Marvell 1.6 ghz CPUs with 512MB RAM. Both have a single USB 2.0 and two USB 3.0 ports. Power consumption is very similar at 12/13 watts (5/6 watts sleep). The key difference is the single bay TS-120 has an eSATA connector. According to QNAP, the TS-120 is middle end and the TS-212p dual bay is an entry level model.

In terms of storage, I slapped a 3TB Seagate on the TS-120 and two 4TB Hitachis on the TS-212p. The TS-212p is running in a RAID mirror for redundancy. Both are similar in performance and capabilities.

Installation is straightforward. Screw in some drives, boot, update the firmware and access everything via a web base interface.

So what do I like about them? They're very much plug-n-play and zero hassle. It is kind of refreshing.
The browser interface acts like a pseudo operating system. In fact, QNAP does call it an OS. QTS 4.0. For a browser interface, it works surprisingly well within the limitations of HTML. You can drag-n-drop files and overal UI is well done for an HTML app. In fact, this works well with Chromebooks.

In addition, QNAP makes some fairly competent iOS and Android client apps to access the NAS.
Importantly, QNAP has a good repository of add-on applications.

You can install apps such as Squid Proxy server, WordPress, Node.js, Python, Plex, an USB web cam, etc. Pretty much, your useful standard open-source applications. Hence, you can turn your NAS into a variety of appliances from downloading bit-torrents, serving DLNA media, to a CMS application server. The capabilities are limited to the computing power of the specific QNAP device you are using. For example, I have Plex installed but neither devices are suitable for real-time transcoding. They serve h.264 MP4 media just fine to my iOS and Android devices but for 1080p MKVs, you'll want to look at the more advance QNAP models.

My needs are pretty simple. I wanted a simple TimeMachine box in addition to storing MP4 media files that I don't want running on the family Plex server.

For these two use cases, the QNAP boxes were fantastic. I even installed the CloudLink add-on which allows you to remotely access your NAS from the outside. Again, it was plug-n-play. There was no need to set-up port forwarding. The Cloud Link creates a tunnel link that allows you to access your NAS via Web browser or a mobile app. Sure, I had this before with ad-hoc set-ups using Pogos and open source cloud tools like OWNcloud/PyDIO. However, none of those solutions worked well. The iOS and Android app has one simple thing I like and that is searching. 

Something so simple yet profoundly compelling. Sure, connecting to a file server is trivial using CIF/SAMBA apps but the ability to search the file-system easily wasn't there. For example, open up ES File Explorer on my Android phone, mount a Samba mount, and I'd always have to browse folders; looking for what I needed. I normally, over-think things but this simple feature is a good reason for me to use QNAP.

There are some handy apps I want to mention real quick. You can set-up an Amazon S3/Glacier rsync. You can even rsync your Google Drive to a path on the NAS. The replication and synchronization tools are pretty good. I was able to set-up an schedule RSYNC to my FreeNAS box and it worked effortlessly. Thus, whatever is copied to my single bay QNAP, a duplicate copy is made in the background to my "real file server." You can even set-up one-touch USB drive copies. If you have an external USB drive, the NAS can easily synchronize to it whenever it is connected to its USB ports.

In terms of performance, it depends on your drives and network. I have a gigabit network and in comparison to my FreeNAS box, the QNAP was considerable slower with mix performances. The Seagate single bay TS-120 was pretty slow in terms of network copies in both AFP/SAMBA.

TS-120 Single Drive Seagate ST3000 3TB drive.

TS-212P w/ Mirror Hitachi 4TB drives

Overall, I'm not really expecting much in terms of performance. These are entry-tier devices. For Time Machine backups and accessing via my iPad, the performance is fine. However, I am genuinely curious about some of their small business models. I am seriously thinking about testing the soho models at my work for TimeMachine backups and RSYNC replication. I'm also seriously thinking about evaluating these for iSCSI vSphere application. 

Conclusion:

I like these devices. I really like the simplicity and ease-of-use. After a full week of work, I don't really want to think about tinkering around anymore on the weekends. I like how they just work for my needs. Between the two, it is a toss-up. I really like the eSATA port on the TS-120 but the TS-212p in a RAID mirror is more disaster-recovery friendly. If you are not sure, I suggest you try the low-end TS-120 if you can get one for $80 on sale. Who knows, you might get hooked. Thats what happened to me. I had no intention of replacing my other devices but the ease-of-use is pretty compelling. At first, I thought of just having a secondary TM back up on my network but now,I am thinking about the 4-6 bay models. I had no intention about getting hooked into a NAS eco-system but I am sure glad I tried.

Links: 

TS-120 https://www.qnap.com/i/en/product/model.php?II=102

TS-212p https://www.qnap.com/i/en/product/model.php?II=117

Drobo 5D update. Now with five 4 TB drives

I decided to upgrade my Drobo 5D; replacing all five 3TB Seagates with five new 4TB Seagates.

I now have 14.5 TB of real useable space; using  4 Terabyte drives. I offloaded my 4 month old 3TB drives into a cheap 4 bay eSATA RAID box. That right there is another 9TB to play with under RAID5.

Just a quick test with Black Magic Speed Test.

Restoring from backup isn't so bad. I'm sustaining and averaging 200 MB/sec from various external STAE129 Thunderbolt drives. This restore consists of mostly small text,photos, and source code files via rsync. The restore is also taking a few hours, so peaking at 200 MB/sec for that long of a period is very good in my book. 3TB worth of data takes about 7 hours or so to restore. I remember that use to take 1-2 days.

SansDigital TOWERRAID TR4UTBPN review

The SansDigital TowerRaid TR4UTBPN is a 4 bay RAID enclosure with eSATA and USB 3.0 interfaces. It was recently on sale so I picked one up.

I've had this sitting around for a few weeks and I've finally freed up some time to play with it.

SanDigital makes over a dozen different enclosures and it can easily get confusing which one to get.
You may have seen some on sale combined with eSATA cards. Take note, many are simply JBOD (Just a bunch of disk) enclosure box that require an eSATA port with port multipliers to function.  Otherwise, you only see the first disk. Hence, the reason many come with eSATA card bundles.  Port multiplication is not common on many eSATA interfaces; including most laptops and motherboards. Those cards often require some sort of software to run them in RAID.
So it is very critical you are aware of the difference when you shop for these type of enclosures.

Luckily, this is not a JBOD enclosure. This has a built in RAID controller and when it was priced the same as a JBOD box, I decided to pick one up. This device supports RAID 0,1,3,5,10 and as well as JBOD. These go for $179-199. I was fortunate to pick this up for $99.

Physical characteristics:


The box is rather tiny and minimal to accommodate 4 drives. They drives do not use a tray but do require you to fasten them down with screws.

The back takes a standard prong power plug (power supply is internal). There are two interfaces - eSATA and full size USB 3.0. There is also a dial and reset button to set the RAID modes. The front side has some little ventilation but I would prefer that the front door provided better airflow.

Setting the RAID mode is a straightforward affair. You simply turn the dial to the mode you want. You can also overcome the dial settings through the RAID manager application.

Front indicator shows activity and which drives are failing. In this case, the 3rd drive has gone bad.

The Raid Manager app shows you status and has other options to create and delete new RAID sets.

The application also provides diagnostic and notifications controls. I've read that the email settings may have some problems with SMTP servers using START/TLS authentication. I did not test to find out. Unfortunately, I could only get the Windows version to run. The Mac app didn't appear to work for me. This may be a Mountain Lion compatibility issue.

USB 3.0

Forget about it. USB 3.0 simply didn't work for me. I've read this complaint on various customer reviews and it appears the JMicron controller has some incompatibility issues with other controllers. I've tried on 4 different machines under OSX, Linux and Windows 7. All had problems with disconnects or simply not mounting. Under OSX, it would disconnect under heavy load or large file transfers. Under windows and Linux, it would randomly disconnect and reconnect every few seconds.  Windows 7 would prompt me to format the unit every 5 seconds.

There may be a firmware update and I'll look into it and update this post later. For now, I would not recommend this as a USB 3.0 RAID enclosure. I've been reading other brands and make have similar problems.Customer reviews on Newegg and Amazon corroborates similar experiences. However, under USB 2.0, I didn't notice any problems.

Speed

Unlike most of the customer reviews I've been reading, I've been getting the 200 MB/sec sequential read transfers in various testing when using eSATA. The drives I used were 500GB 7200 rpm Seagates.

Windows appears to perform better (using benchmarks as well as real copies).

I only tested RAID 0 and RAID 5 but I did not see much of a performance loss going to RAID 5. The 4k/512k was about 1/4 slower  in writes but the sequential read/writes were similar to RAID 0. In short,  copying large files saw little loss while your random files will be faster under RAID 0 as it should be.

Here is a RAID 5:

and RAID 0

RAID 0 under Linux

Conclusion.

This box is cheap enough to get if you are using it with eSATA. I would not recommend for USB 3.0 use.

If you are a laptop user without eSATA, I recommend using a USB 3.0 eSATA adapter like this NewerTech one I reviewed earlier.

Link: http://www.sansdigital.com/towerraid-plus/tr4utbpn.html

New toy. $100 RAID enclosure.

I haven't had much time to post. Once I get drives, I'll see how this $100 RAID box compares with a $850 drobo.
Initial testing with old 500gb drives look promising.

Drobo 5D USB 3 and Thunderbolt Review

I've had some time to play with the Drobo 5D; filling up about 6TB worth of data (out of 10.89TB in my 5X3TB setup). I've been using it in a mix-use environment. I think I have enough time to form an opinion. So here is my non-professional, regular end-user hands on review of the new Drobo 5D Thunderbolt / USB 3.0 raid box.

Introduction

The device was originally announced during the summer and there was a series of delay, waiting and anticipation. Drobo did not release performance figures until they neared shipping. This was not a very promising introduction.

I was ready to give up and go elsewhere. The Pegasus R4 is the closest competitor to the Drobo and I'm sure it will be cross-shopped. The Pegasus R4 (4 bay RAID) system can be had for $1,000 with 4TB of disks ($985 at Provantage link, $1100 amazon link).

The Drobo 5D is $850 and it comes bare (enclosure-only) but does include a Thunderbolt cable (which is worth $50 by itself). You will need to load it up with drives.  If you are a savvy shopper, you can start off with three 2TB drives to give you 3.63TB of useable data (according to the online calculator) to match the Pegasus R4's storage capacity. So if you shop around, three 2TB will cost about an $300 (with heavy slickdeals-like discount and bargain hunting) which still makes it more expensive than an R4 setup.

Even though the Drobo 5D is more expensive (with drives), you do have 5 bays vs R4 only having 4 bays. You also have the added benefit of USB 3.0 which the R4 lacks. This broadens the use of this device over a Thunderbolt only system. With 5 bays, you have the potential of filling it up with 4TB drives; giving you around 14.5TB useable storage (from 16TB). That in itself is very compelling.

You can start off cheaper with a Drobo because you can add drives at any time. But, I doubt anyone buying a Drobo at $850 will skimp out by using 1 or 2 drives.

So why would I go out and buy this device? Well, I was already allotted a budget and "gifted" funds to buy one from a job I did over the summer. In short, I had to buy one or return the bonus funds I got from that job.

The Drobo Sales Pitch
Drobo claims this device will yield 400 MB/s reads with 256 MB/s writes on average. That doesn't sound as compelling as their competition who posts higher Megabytes per second. Rather, Drobo prefers to bench based on I/O per seconds (IOPS). They believe that "real-world" performance is where the 5D will shine.

They explain this in detail on their performance page: http://www.drobo.com/how-it-works/performance-professional.php

In short, they claim:

"While benchmarking give you a good indication of how well a storage device performs when it’s healthy and for a single task, it is still not a replacement for real-world performance. Benchmarks are nice, but real-world performance is everything."

Real-world performance includes a mix workload; copying various random small files in addition to large bulky files. This is where the IOPs figure comes into play. If you are copying a Photo Library folder with lots of small thumbnails or large volumes of small text files (say a web project), your transfers will be slower than if you normally copy something like large video files. I've seen this happen in the real world where I copied GIT repo of a website with thousands of files and it did indeed take longer than copying large movie files.

They claim 400 to 1280 IOPS.  For comparison, some single SSDs are rated at 40,000 to 80,000 IOPS. The Samsung 830 SSD has benched 4K random read IOPS of 80,000.

Other HDD manufacturers do not bench on IOPS and I could not find any reviews on the net to compare the Drobo 5D to when it comes to comparable storage/raid redundancy.

If you are used to thinking in Megabytes per second, you need to acclimate yourself to IOPS.

The Drobo uses a form of RAID which is similar to RAID5/RAID6/RAIDZ. They call it Beyond Raid and it has a some unique characteristics both real and perceived. It provides single or dual disk redundancy. This in itself is very important. This is a redundant RAID system.

Unlike traditional RAIDs, you can mix-and-match drives sizes (called Mix drive utilization). You can also grow your array at any time buy swapping out smaller drives for larger ones. You can start off with 1TB drive, add 2TB in a few months, or opt to mix and match with 3 and 4TB drives as prices drop. This provides un-paralleled flexibility.

Unlike other RAID systems, Drobo implements their version of data-tiering. Tiering, not to be confused with SSD caching, basically will automatically stores the files that require faster access on SSDs. The system automatically tunes itself and migrate data across SSD/HDD depending on usage. This is similar to the "Fusion" drive technology that Apple recently announced for their new iMacs/Mac Minis. Like the Apple's Fusion drive, the Drobo requires an SSD in the form of a mSATA drive that is installed in the bottom bay. This is the key to increasing IOPS performance.

Their pitch on data-aware tiering can be found here: http://www.drobo.com/how-it-works/data-aware-tiering.php

Using SSDs for acceleration, smaller and random reads benefit significantly while the larger HDDs are used for storing larger data.  This whole process is automated. Transactional tiered data goes to SSD and bulk data goes to platter drives. I can see the benefits for databases and virtual machines.

For a non-technical person or someone unfamiliar with RAID technologies, the Drobo 5D is a simple and easy to use system. In theory, you can't go wrong. You won't get the fastest drive setup but you will get ease-of-use and a rich feature-set. Automated data-tiering is a very compelling proposition.

The things I like

• Battery backed cache. This is found in more expensive RAIDs and helpful in case of power failure.
• The software has a nice GUI interface with tools like email notifications.
• Two Thunderbolt ports for daisy-chaining and it does indeed work well.
• Black thunderbolt cable that is slimmer than Apple's cable. Where can I buy another?
• Ability to mix drives and upgrade in the future.
• Thin Provisioning.
• Ability to boot from Drobo.

The Competition

Drobo's competitors here are other DAS (Direct Attach Storage) units rather than NAS like  the Synology.

As I wrote earlier, the Pegasus R4/R6 are the closest competitors to the Drobo 5D. The Pegasus  R4/R6 yields  635 MB/Sec read 535 MB/sec writes. R6 averages both 680 MB/sec read/writes.

You can get up to 1GB /sec when you load them with SSDs. Impressive indeed.

Reading Links:

http://www.anandtech.com/show/4489/promise-pegasus-r6-mac-thunderbolt-review/6

http://www.tomshardware.com/reviews/my-book-thunderbolt-duo-pegasus-r4-2big,3222-8.html

The Pegasus units can be configured in different RAID types to suit your needs.

There are also other compelling Thunderbolt options in the market. Even the LaCie Big Disks goes up to 635/MB sec  (link: http://www.lacie.com/products/product.htm?id=10549) And it start at $399.

The Western Digital My Book VelociRaptor Duo has 1549.77 IOps but that is a dual drive RAID 0 Thunderbolt enclosure (http://www.storagereview.com/western_digital_my_book_velociraptor_duo_review).

In my opinion, the HDD based Thunderbolt system yields the biggest bang for your buck. You get better than eSATA/USB 3 and in some cases near or exceeding SSD territory for modest money.You can't buy 10TB worth of SSDs or a SAS based RAID cheaply. This is why the Drobo comes in. You definitely have the ability to have lots of storage.

Testing

Let me preface by saying I am NOT a professional journalist or reviewer. I am enthusiast end-user and a normal consumer like you.

My source drive is a Crucial M4 512GB SATA 6.0Gb/s running on a Macbook with Mountain Lion 10.8.2. The Crucial M4 random reads bench at 45-50,000 IOPS (http://techreport.com/review/20646/crucial-m4-solid-state-drive).

In short, my source copies will not be bottlenecked copying to the Drobo.

My Drobo 5D setup:

• Five Seagate ST3000DM001 3TB 7200 RPM 64MB Cache SATA 6.0Gb/s 
• Crucial M4  mSATA 6gb/s 128GB SSD w/ latest firmware 01MG as the cache accelerator drive

My set-up gives me 10.89TB useable space (15TB of drive, actual size 13.64TB).


Knowing that the Drobo 5D will not bench that great using MB/sec, I attempted to evaluate this device on Drobo's premise of mix-real world usage. However, I will also present synthetic Black Magic numbers for reference.

In terms of USB 3.0 performance, this is what Black Magic shows. 90MB/s write 110-130MB/s reads.

I've gotten better performance with eSATA boxes like the OWC Elite Pro Qx2 connected via USB3-to-eSATA adapters. It would be unfair to compare this number to faster performing single drives because this is a redundant RAID set-up. However, the OWC Elite Pro Qx2 would also be in a RAID5 format so these numbers are not that great in terms of Black Magic benchmarking.

Next, Thunderbolt. the best I could muster over several tries were 210-240 MB/s writes and 290-300MB/s reads. This is actually good for a consumer RAID but fall short when compared to something like the Pegasus R4.

Again, these are just synthetic numbers that an application produces and is not indicative of real-world use. Im not a professional reviewer so I can't give you IOPs numbers.

To follow Drobo's assumption of real-world mix usage, I attempted to see how well it performed copying a mix of different file types.

Instead of trying another synthetic test, I did it the old fashion way. I timed it. My test is all writes. I am more concerned about fast the drive writes my copy rather than how fast it reads.

Instead of copying large 30GB files to see the sustained transfer rate, I opted to try a typical usage of copying my photo backups. This s a 28.76GB folder of 2,277 photos in RAW format from my DSLR. This is 2-3 months of photos and typical usage example to test performance of copying a lot of medium sized files where the should Drobo shines.

I called this the Photo test.

I did another test with extremely small files. A webserver folder with 27,759 items clocking it at 1.5GB. These include small 5k HTML/Javascript files. So this would be a typical thing for a web developer to back up his/her files of a working project.

This is the WebTest.

I compared it with a bunch of different drives to get an idea how the Drobo fared. Obviously, it won't be as fast as SSDs but I tested with some SSDs just for reference.

• Samsung 830 SSD via USB 3.0 in an Oyen Mini-Pro enclosure (with ASMedia 1051e chipset).
• Samsung 830 SSD w/ Seagate Thunderbolt Adapter
• Internal Seagate Momentum XT 2.5 " 7200 rpm drive. This is the Hybrid XT drive.
• 5400 rpm USB 3.0 Seagate Go Flex portable drive
• 5400 rpm Firewire 800 Seagate Go Flex portable drive
• USB 3.0 Hitachi portable 2TB drive. 5900 rpm.
• HDD Thunderbolt comparison against a LaCie 2Big 4TB in RAID 0.

Here are the results... I will give a summary and interpretation below.
Photo Test. Copying 28GB of Photos.

Shorter bars are always better.

Web Test. Copying a folder of HTML/Javascript/PHP files.

Full Chart.

	Device	Time (min.sec)
Photo Test	USB3.0-Samsung SSD	1.54
Photo Test	Thunderbolt-Samsung SSD	1.30
Photo Test	HDD Seagate 7200 rpm internal drive	6.18
Photo Test	5400 rpm 2.5 Go Flex USB 3	6.18
Photo Test	5400 rpm 2.5 Go Flex via Firewire 800	6.18
Photo Test	5900 rpm external Hitachi USB 3	10.02
Photo Test	LaCie 2 Big Thunderbolt 4TB	1.47
Photo Test	Drobo 5D	3.58
Web Copy	USB3.0-Samsung SSD	2.28
Web Copy	Thunderbolt-Samsung SSD	.27
Web Copy	HDD Seagate 7200 rpm internal drive	.41
Web Copy	5400 rpm 2.5 Go Flex USB 3	12.04
Web Copy	5400 rpm 2.5 Go Flex via Firewire 800	1.36
Web Copy	5900 rpm external Hitachi USB 3	14.41
Web Copy	LaCie 2 Big Thunderbolt 4TB	.28
Web Copy	Drobo 5D	1.04

Test observations.

Going in,  I knew the SSDs will smoke any HDD system. That is a given.
The Drobo 5D performs decent in my view. It is not super fast and not slow. The LaCie 2big Thunderbolt simply smokes it but we have to remember, the LaCie is a Striped RAID vs Redundant Array. Also, the LaCie is  4TB/6TB  vs 10TB (you can go up to 16TB on the Drobo).

If you notice, Firewire smokes the USB implementation when it came to small 4k-10K file web copies. This illustrates the inefficiencies of the BOT (Bulk-Only-Transport) protocol of USB 3.0.

Watching the progress bar on a regular USB 3.0 drive was painful to watch as it struggles with smaller files. 12-15 minutes for a 1.5GB folder!

The Drobo 5D performed surprisingly well. Like I said earlier, the synthetic benchmarks will not give you a realistic idea of how the drives will perform in the real world.

So is the Drobo 5D fast?

I wish I had a Pegasus R4 to compare. I ran this test about 3 times on fresh formatted systems. I also skipped testing the Drobo 5D via USB 3. It simply wasn't that fast with USB 3.0 in my view.

It did surprisingly well where it was suppose to -  with smaller files. If you are copying 4-6GB Blu-Ray ripped movies, there will be other things out there that will out-run it.

How it fares will ultimately depend on how you plan to use the device.  I would also add that my test may not be indicative of your use. The device is suppose to automatically tier the data upon repeated use and you might fare better in a different workload environment.

The speed perception will also depend on what you are used to. I've been living with SSDs in my day-to-day life that I'm very spoiled. I carry about 4-5 SSDs in my carry bag. The LaCie 2big is also a very fast HDD Thunderbolt enclosure. Read and writes average 300+ MB/sec. But I have to remind myself that the LaCie is a striped RAID and not a redundant RAID like the Drobo 5D. The redundant RAIDs I've used are the OWC Mercury, MediaSonic, and Sans-Digital eSATA based systems for consumers. At RAID5, they often top out at 150-200 MB/sec. So when you compare it to the Drobo 5D, the Drobo is indeed faster. However, those enclosures are  $300-400 versus a $850 Drobo.

Is this for you?

First and foremost, you need to have a Thunderbolt equipped machine to be in the market for this device. The USB 3.0 is a nice add on for future proofing but it isn't the main reason to buy this device.

Should you use this device in a network share environment? No. This is not a NAS or even recommended as storage for your SOHO server. I've read of people planning to use this device ( and other Thunderbolt RAID systems) as storage for mac mini server set-ups. With Gigabit's theoretical 125 MB/sec limits, this device is over-kill on single NIC servers. You'll need Fast multiple NIC teaming or 10GbE networks to properly use this for multiple clients.  Otherwise, there are mini-SAS eSATA solutions. Look elsewhere for a dedicated NAS system for your SOHO use.

There will be users who buy the Drobo 5D as their primary backup system. Like with all backups, it is not a good idea to have all your eggs in one basket because with one catastrophic failure, you lose everything. There are countless horror stories on RAID failures. Drobo is not unique . If you plan to use this as your primary backup, please consider having additional offline or additional supplemental backups as well.

I think the buyers will most likely be creative professionals who need fast, large storage. They include photographers and video editors who need to move large files fastly. If you are planning to copy and archive  your MP3 music collection, this is definitely over-kill.

My use case would be a mix usage with a heavy leaning on using it as a fast work drive. E.G. scratch drive, running Virtual Machines, Lightroom catalogs.

Drobo with other platforms.

Right now, Macintoshes have a monopoly in terms of Thunderbolt usage. I'll follow up in the future to give my opinion on using this with Linux (via USB 3.0).

Conclusion

You need to heavily weigh your needs and your usage to consider this device.

Filling up a Drobo is not cheap. 5x 2TB drives would cost you around $500 (if you can get the drives at $100 a piece) and this would yield you 7.26TB of usable space according to their drive calculator.
Dont even bother with 1TB drives. The difference being $70 vs $100 2TB on sale, go for the 2TB or 3TB HDD drives. Hence, we are now talking about around $1350 to begin. This does not include the mSATA cache accelerator drive.  With a 128GB mSATA cache drive, you are looking at $1500.
$1500 can buy you lots of drives, SSDs, and even a few RAID enclosures.

As for a redundant backup system, you may want to skip the Drobo 5D because I do believe the $850 is a bit pricey for that type of use. If you absolutely need to have something faster than your current backup system, then the Drobo 5D may be an option for you.

As for a performance oriented work drive, it is a tough call. The Pegasus R4 is a proven performer with little problems (according to end users and message boards). You can buy an R4 for your fast RAID5 and a few external USB 3.0 3TB drives for your secondary backups. Or you can buy a LaCie Big Disk (I got a 4TB refurb for $200) and a lot of spare 3-4TB external drives.

The picture below exemplifies this hard choice. $900 can get you a LaCie 2big, 2 SSDs, three 3TB USB 3.0 external drives (totaling 9TB) vs an empty bare Drobo 5D enclosure (with no drives). 13TB vs 0 on the Drobo.

For me, it boils down to this, how big of a scratch work drive do I need? If you need 6 TB and larger, the Drobo is a good buy.

There is no substitute for the fact you can get up to 16TB of storage with good speed. 200-350 MB/s is still very good.

If you are looking for a combination of both in terms of mix-use (backup drive and work drive), then you might buy the sales pitch Drobo is making.

As for me, I think a mix-setup that I currently have is more preferable: A fast 4TB cheap Thunderbolt striped RAID or SSDs for work disk and a NAS for backup along with multiple external drives/cheaper RAID setups.

If the Drobo 5D was $100-$200 cheaper, I would totally recommend it.

I'll probably end up keeping this device. I can always use the storage.

Update 2012-12-03: Thunderbolt does not work under Windows. Read more on my blog post.

La Cie 2big thunderbolt enclosure w/ 2 SSDs

Just in. I got a La Cie thunderbolt 2 big drive and loaded them up with some Samsung SSDs.
400 MB/s write and 475 MB/s reads.

Samsung 830 SSD USB 3.0 RAID-0 in OSX and Ubuntu 12.04

Striping two USB sticks was cool but I wanted to try it further with two fast SSDs.
As many people know, the Samsung 830s are fast, reliable and great SSDs. They make great candidates for a SSD based RAID.

In my search for a super fast consumer grade RAID solution, I figure I try USB 3.0 and see what happens next.

I did tests on both Mac OSX Mountain Lion and Ubuntu 12.04 (off a Gigabyte GA-H67N-USB3-B3 motherboard).

On the Mac. Making a stripe software raid-0 is pretty straightforward using Disk Utility.

In Ubuntu 12.04. Setup is almost as easy as the mac.

The results were not that great as the Thunderbolt solutions I've seen. In fact, I now think it is is more economical to get platter based spindle thunderbolt RAID drives instead of trying USB/eSATA with SSDs.

I'm getting 200 MB/s writes on both OSX/Linux and 200-300MB/s reads. Have in mind, these are blazing speeds if you only used to using regular HDDs. However, the results are unimpressive in a RAID-0 array.

Real world copy. 200 MB/s

In fact, I see no real speed gains striping two SSDs with USB 3.0. Single SSDs are just as fast.
Or simply, the USB 3.0 controllers are not up to speed with the fastest SSDs yet. This may explain why I haven't seen any interesting or worthwhile USB 3.0 RAID enclosures on the market.

I think I'm going to have to spend the cash on a Thunderbolt solution because eSATA and USB 3.0 is not cutting it for a DAS (Direct Attach Storage) solution.