[EricTheRed]

I'd like to find out if anybody offers an eSATA version of something like a Drobo that I can use with time machine to back up but is fast enough not to lock up the OS while it connects and spins up the external HDDs.

Ready and willing to be clued in.

[zwiebel_]

This may be something to consider
Mini Storage Tower from Addonics, part #MST5XHPM (comes with a hardware port multiplier).

[OreoCookie]

Not really, the drobo is one-of-a-kind. But I'm sure it's a popular request, be sure to drop them an e-mail.

[mduell]

It wouldn't matter; it's not the interface that holds Drobo back, it's the internal processor.

The solution I've found (and will implement for myself pretty soon) is a pair of 5 drive RAID0 enclosures with a software mirroring (either softRAID1 or rsync/similar every $time-period... I prefer the latter). It comes out very cost competitive with Drobo, much faster, and almost as easy to expand with a much lower risk of hosing your data while expanding (or the ages-long rebuild process).

[Mastrap]

My Drobo, latest firmware applied, has yet to lock my system up. I've got it stocked with two 1TB drives and there is no noticeable lag.
I understand that the Drobo is a consumer product, but for somebody like me who really can't be asked to set up a RAID configuration at home it's the perfect solution.

Originally Posted by mduell 

with a much lower risk of hosing your data while expanding (or the ages-long rebuild process).

I would like to know how you arrived at this conclusion. RAID arrays are sensitive to all sorts of issues, voltage spikes, incompatible drives, you name it. I am not saying that they are in any way inferior to the Drobo, but they are certainly far from perfect. For anybody without a considerable amount of technical knowledge it is far easier to hose a RAID system that it is to hose a Drobo.

The rebuild only takes place, ever, when you remove drives while the unit is live, i.e when you're hot swapping drives. Except for once demonstrating this functionality to a friend, I've never had the need for the Drobo to rebuild anything. Also, throughout the rebuilding process your data is readily available to you so I fail to see why this should be an issue.

[nonhuman]

Originally Posted by mduell 

It wouldn't matter; it's not the interface that holds Drobo back, it's the internal processor.

The solution I've found (and will implement for myself pretty soon) is a pair of 5 drive RAID0 enclosures with a software mirroring (either softRAID1 or rsync/similar every $time-period... I prefer the latter). It comes out very cost competitive with Drobo, much faster, and almost as easy to expand with a much lower risk of hosing your data while expanding (or the ages-long rebuild process).

Have you considered using ZFS instead of RAID? I think that's what I'm probably going to be doing very soon as I'm very soon going to outgrow my current RAID setup. I haven't started playing with it yet (it's actually a big part of the reason I started fooling around with FreeBSD over Linux), but from what I've read it seems like it should be a much simpler method of accomplishing the same result.

[mduell]

Originally Posted by Mastrap 

My Drobo, latest firmware applied, has yet to lock my system up. I've got it stocked with two 1TB drives and there is no noticeable lag.
I understand that the Drobo is a consumer product, but for somebody like me who really can't be asked to set up a RAID configuration at home it's the perfect solution.

With two drives you've turned your $500 enclosure into a $120 enclosure, but slower. The latter has no more setup required than the Drobo; less, even, since you don't need to install the Drobo dashboard to see how much space is actually available.

Originally Posted by Mastrap 

Originally Posted by mduell

with a much lower risk of hosing your data while expanding (or the ages-long rebuild process).

I would like to know how you arrived at this conclusion. RAID arrays are sensitive to all sorts of issues, voltage spikes, incompatible drives, you name it. I am not saying that they are in any way inferior to the Drobo, but they are certainly far from perfect. For anybody without a considerable amount of technical knowledge it is far easier to hose a RAID system that it is to hose a Drobo.

Anywhere a pair of RAID0 enclosure is susceptible to a voltage spike (external or internal to the enclosure), Drobo is also susceptible. I don't know what you mean by incompatible drives; of course you can't mix IDE and SAS with my solution, but Drobo doesn't support either anyway. You only need to poke around DroboSpace for a few minutes to find the "Drobo started a relayout, now my data is a turnip" posts. With a pair of enclosures you always have a "real" copy of your data sitting around somewhere during an upgrade; I understand that the theoretical underpinnings of parity are sound and well known, but the implementations leave much to be desired.

Originally Posted by Mastrap 

The rebuild only takes place, ever, when you remove drives while the unit is live, i.e when you're hot swapping drives. Except for once demonstrating this functionality to a friend, I've never had the need for the Drobo to rebuild anything. Also, throughout the rebuilding process your data is readily available to you so I fail to see why this should be an issue.

A relayout takes place every time you add a drive (assuming your existing array is more than half full). Yes, the data is still available, but it goes from "not fast" to "agonizingly slow" during the relayout process. My solution also leaves the data available at degraded performance (still better than Drobo can hope to be) during a capacity upgrade.

Originally Posted by nonhuman 

Have you considered using ZFS instead of RAID? I think that's what I'm probably going to be doing very soon as I'm very soon going to outgrow my current RAID setup. I haven't started playing with it yet (it's actually a big part of the reason I started fooling around with FreeBSD over Linux), but from what I've read it seems like it should be a much simpler method of accomplishing the same result.

ZFS is a great tool, but for my usage it causes more problems than it solves at the moment. I expect that I'll implement it or something similar in the future when there are widespread, robust implementations.

[Mastrap]

Originally Posted by mduell 

With two drives you've turned your $500 enclosure into a $120 enclosure, but slower. The latter has no more setup required than the Drobo; less, even, since you don't need to install the Drobo dashboard to see how much space is actually available.

Anywhere a pair of RAID0 enclosure is susceptible to a voltage spike (external or internal to the enclosure), Drobo is also susceptible. I don't know what you mean by incompatible drives; of course you can't mix IDE and SAS with my solution, but Drobo doesn't support either anyway. You only need to poke around DroboSpace for a few minutes to find the "Drobo started a relayout, now my data is a turnip" posts. With a pair of enclosures you always have a "real" copy of your data sitting around somewhere during an upgrade; I understand that the theoretical underpinnings of parity are sound and well known, but the implementations leave much to be desired.

A relayout takes place every time you add a drive (assuming your existing array is more than half full). Yes, the data is still available, but it goes from "not fast" to "agonizingly slow" during the relayout process. My solution also leaves the data available at degraded performance (still better than Drobo can hope to be) during a capacity upgrade

Data loss happens. It can happen with a Drobo, it can happen with a RAID array, it can happen with any drive based storage solution. There is no data that I have seen that suggests that the Drobo is in any way less reliable than any other backup solution.

As I've said, there is no reason why anybody should wish to add or remove drives from a Drobo with any regularity - it just doesn't happen.

Drobo is constantly updated, with new firmware addressing issues as they are discovered. If you have experience setting up, and supporting, a RAID setup then that's the solution to go with, but for the home user who wants a solution that just work the Drobo is a decent solution.
As far as cost is concerned, this is an interesting article:

At Data Robotics, I am asked many variations of these questions:

Why shouldn’t I buy a “fixed capacity RAID box” (one that doesn’t let you expand its capacity) now and then buy another when it gets full?
Why shouldn’t I buy an “expandable RAID box” that lets you manually replace each and every drive with a larger one in order to increase its capacity?
Isn’t buying an empty Drobo and then some disks more expensive than getting RAID?
Would you be surprised to learn that RAID systems will always cost more than Drobo? This results from the combination of RAID’s limitations and the pricing dynamics of disk drive makers. Continue reading to find out why.

RAID’s Economics: You Pay Dearly Today For The Future Growth

One problem with RAID systems is that they force you to buy more capacity today than you need. Because adding storage capacity is either impossible, or very difficult and labor intensive, RAID buyers purchase more than they need and forecast how much they will require in the future.

Historical trends show hard disks decreases in cost 40%/year. This means RAID systems waste your money. You have to spend money today on excess storage capacity – storage that you could buy for much less if only there were a way to postpone purchasing it.

And to make matters worse, with most RAID systems its impossible to increase storage capacity. When they are full, it is time to buy another box. Very few RAID boxes will let you increase capacity. With those that can, you must upgrade all drives in the system, one-at-a-a-time, before more capacity is available. This is a labor intensive process – you have to remove a drive and replace it with a larger one. Because this “breaks” the RAID set you have to wait for the entire RAID array to rebuild before the next drive can be replaced. This typically takes 4-6 hours per drive not counting assembly time. For a typical desktop RAID array, this means 2-4 days elapsed time.

Isn’t there a better way?

Drobo’s Economics: You Save Money and Pay As You Grow

One of Drobo’s design goals was to let users upgrade storage forever. As long as larger drives are available (and they have been for the past 30 years) just insert a larger SATA drive from any manufacturer to increase Drobo’s storage. This empowers users to select drives with capacity and price from manufacturers they find attractive. Drobo lets users harness market pricing dynamics for their own benefit. Because hard drives are continually getting denser it is advantageous to postpone purchasing a disk drive until it is actually needed.

Bird’s Eye View of Hard Disk Economics

Even though hard drives increase in capacity 40% per year, their manufacturers can not increase their prices due to market competition. Each vendor usually has three or four price “bands” in which it sells drives.

The premium price is for their latest capacity drive, and the price band is approximately $300-400.
The “value” price is usually around $100.
Depending on the vendor there may be one or two intermediate price bands, around $200 or $300.
Over a drive’s produce life cycle it will start at the premium price band, and move downward as newer, larger capacity drives are introduced. After 18 months a drive usually goes out of production.

Because drive prices constantly decrease it’s un-economical to buy RAID systems. It’s like setting your money on fire.

Putting it Together: Watch Your Dollars

DRI’s research shows that the average prosumer considering Drobo has 702GB of data and that it is growing at a median rate of 72% per year. This may sound unbelievable, but consider all the digital media we are creating and collecting like photographs, amateur and professional video, music, and even work files.

What it will cost an average prosumer to store that data, as in increases, over the next few years? Lets consider three alternatives, comparing two RAID boxes with Drobo:

Buying a fixed capacity RAID box with enough capacity to last at least one year. When more capacity is needed, another box will be purchased. A well known, typical four drive desktop RAID was used for comparison. A recent price check at Amazon showed this cost $1,099 before tax & shipping. We assume the future product is purchased at the same price point.

Buying an expandable RAID box with enough capacity to last at least one year. A four drive, desktop system with a network interface was used for comparison. A recent price check on Amazon for a system with 1,500GB of protected capacity cost $1,380. When more capacity is needed a set of four additional drives is added.

Buying a Drobo: a Drobo is purchased with enough protected storage to meet the user’s immediate needs. When more capacity is needed, another drive is inserted, or a smaller drive is replaced with a larger one. Initial acquisition is $799 (i.e. $499 for Drobo plus (3) 500GB “value drives” purchased separately.)
Data Growth and Capacity Required

The chart below shows how the user’s storage requirement will change over time. Notice that with Drobo it is possible to closely match the amount of storage purchased to that actually required. For the RAID systems, there is considerable over-buying where RAID forces storage capacity to be purchased which is not needed at that time by the user.

[/QUOTE]

[OreoCookie]

I think you're missing the unique selling point of the drobo, mduell. You can pop drives in and out, mix different sizes at your leisure. That's something your RAID solution can't. Especially the more average, less tech-inclined consumer can appreciate the simplicity -- and many are very happy to pay for that.

[mduell]

Originally Posted by OreoCookie 

I think you're missing the unique selling point of the drobo, mduell. You can pop drives in and out, mix different sizes at your leisure. That's something your RAID solution can't. Especially the more average, less tech-inclined consumer can appreciate the simplicity -- and many are very happy to pay for that.

Funny, Mastrap asserts "there is no reason why anybody should wish to add or remove drives from a Drobo with any regularity."

I have yet to meet anyone whose needs are better met (simplicity, availability, performance, cost) by Drobo instead of a RAID based solution. I think the figures in that Data Robotics blog post are bogus and I'll do the math myself tonight.

[mduell]

For everyone playing the home game, here's the url for the massive copy/paste above.

For Drobo both DR and I come up with a 4 year lifecycle cost of $2100 ($800 initial, 13 $100 upgrades) using the chart in the above post.

For my suggestion above, I come up with a 4 year lifecycle cost of $1611 ($900 initial, $313 upgrade at 1.5 yr, $165 upgrade at 3 yr, $113 upgrade at 3.75 yr, and $77 upgrade at 4.5 yr).

Methodology: Use current 1TB ($230)/enclosure ($218) prices, discounted future drive purchases at 12% per quarter (40% per year, as the DR guy asserts).

So I'm ahead by $500 (~25%), have an order of magnitude better performance (~100-150MBps vs ~10-15MBps [remember the Drobo is always >75% capacity]), always have a 'real' copy of my data on hand, and the upgrade process is "add disk, destroy array, create array, copy data."

[OreoCookie]

Again, you forget that you compare Apples and oranges here. Hardware is cheap compared to labor. Setting up a RAID requires considerably more expertise than setting up a Drobo. A Drobo is much easier to set up and can cope with the loss of a drive (if you have sufficiently many drives installed) -- unlike your RAID0. Of course, you can switch back to your other RAID0, but you will have lost some data. The selling point is not performance either, it's supposed to be a backup drive for your pictures and videos, for instance. And it is not supposed to compete with a real RAID solution either. Plus, it's hardware independent and people can carry their data -- that's also something you cannot do with your software RAID0.

No offense, but you seem like a pc hardware guy who buys everything as cheap as possible, forgetting that there is more to hardware and software than specs you print on boxes. If you work with computers professionally, people are usually more expensive than hardware.

[Mastrap]

That's exactly my impression. If you have the expertise, and are willing to put up with the restrictions of RAID0, then good for you. If you want an out of the box solution, and are willing to put up with the restrictions of a drobo, then the drobo is for you. Simple.

[OreoCookie]

Even then, a Drobo and a RAID0 won't satisfy the same needs. I'd be happy with a Drobo as a backup drive, knowing that I never have to manually migrate data and that I have a failsafe just in case one harddrive breaks. It cleans up clutter. If I had the money, of course, I'd prefer a RAID5 box under my desk, but that costs about two to three times as much as a loaded Drobo.

[mduell]

Originally Posted by OreoCookie 

Again, you forget that you compare Apples and oranges here. Hardware is cheap compared to labor. Setting up a RAID requires considerably more expertise than setting up a Drobo. A Drobo is much easier to set up and can cope with the loss of a drive (if you have sufficiently many drives installed) -- unlike your RAID0. Of course, you can switch back to your other RAID0, but you will have lost some data. The selling point is not performance either, it's supposed to be a backup drive for your pictures and videos, for instance. And it is not supposed to compete with a real RAID solution either. Plus, it's hardware independent and people can carry their data -- that's also something you cannot do with your software RAID0.

No offense, but you seem like a pc hardware guy who buys everything as cheap as possible, forgetting that there is more to hardware and software than specs you print on boxes. If you work with computers professionally, people are usually more expensive than hardware.

Given that we answer questions here in our spare time for free, the opportunity cost arguments are somewhat difficult.

I understand that performance isn't on the minds of a lot of users, but when you're measuring your file copy times in hours to days it starts to become more relevant.

I don't see any overlap between Drobo and RAID5; the latter is rarely worthwhile unless you're spending more on the controller alone than the lifecycle cost of the Drobo.

One clarification is that my solution is at least hardware RAID0 and hardware RAID10 when you're looking in the same capacity range as the Drobo. Another solution that I don't think I've mentioned in this thread is 2 drive hardware RAID1 for lower capacities (capacities that I've seen Drobo used for).

I'm well aware of the tradeoff between hardware cost and personnel costs. There are plenty of situations where I'll throw more hardware at a problem, and plenty where I can see it's wiser to throw people at them (DBAs, for example, can be worth their weight in gold).

[OreoCookie]

I certainly wouldn't recommend the Drobo to anyone who needs cutting-edge performance. After all, it's limited by USB2 and the bus is saturated easily by a single modern harddrive. It wouldn't hold a candle to your RAID10 solution, that's for sure.

I definitely see the appeal of the Drobo and a niche for it in the market. By no means it's a one-size-fits-all solution, though

[mduell]

Originally Posted by OreoCookie 

I certainly wouldn't recommend the Drobo to anyone who needs cutting-edge performance. After all, it's limited by USB2 and the bus is saturated easily by a single modern harddrive.

USB2 is not limiting the performance of Drobo; USB2 single drives easily push 25MBps sustained, while Drobo averages about 10. It's the internal processor that does the parity math that limits Drobo performance.

[Mastrap]

Thing is, for many applications this is just fine. I keep my work files on our Drobo, and even my larger InDesign files are rarely larger than 10MB. Perfect. iTunes works without a hitch, so does Aperture, where my RAW files weigh in at about 10-15MB a piece.

If you're frequently shoving huge files about you'll notice the difference, but for the work I do it's simply not an issue.

[tridentinecanon]

mduell talked me out of Drobo about a month ago at another thread, and I'm glad he did. Saved me a heap of cash.

I find the RAID setup to be quite simple and understandable, and I'm no techie. It actually was more promising than the Drobo because with Drobo I would need to buy two of them to accomplish what I needed, which would in many ways defeat the purposes of Drobo.

To tie all this in with the OP's original question, I'm very pleased with all of the products I have ever purchased form cooldrives.com, particularly their eSATA RAID products.