WD Caviar Black are consumer-class drives, as are Hitachi Deskstar.
Drobo’s support philosophy seems more about dropping customers than meeting commitments. When convenient, (16TB Drobos, consumer-class drives in DroboPro), they’ll silently revise their promises and leave their customers holding the bag.
P.S. Spiney, that’s a good point. In this case you can share the experience by dumping flaming horseshit on your head while repeating “Drobo hates me and my consumer-class drives.”
ive always used samsung hdd’s in my drobo pro. they are both green and 5400rpm and a consumer products. ive found them reliable in computers, and ive found them to be reliable in my drobopro. i dont think i would gain anything by replacing them with 7200rpm drives.
I’m wondering if this has something to do with my problem…
I’ve just moved a client’s email database onto a DroboPro to test and it contains 8 WD Caviar Green drives. When I connect the box via iSCSI (thanks Docchris), I get constant activity lights on the DroboPro. I know that the DroboPro does some kind of internal housekeeping and that the Mail Server uses lots of little files in the Store directory.
Server performance seems to have degraded since I made this change and it used to fly along on the internal drives (5 x 146GB SAS HP modules). I had documented 80MB/sec reads and writes to the Drobo since I made the change to iSCSI and thought that this would be a perfect solution for an email database.
paraprod: Mail servers, as you say, do many small file transactions. Further, they’re extremely sensitive to latency (per RFC, “250 OK” should be issued only after the message has been committed to nonvolatile storage).
Unfortunately, Drobo products are almost the opposite of what you need on a mail server.
When I managed high-capacity mail relays (using Postfix on Linux), they were heavily tuned for performance both in software and hardware. It doesn’t have to cost a lot of money, but you absolutely need low-latency file operations and fast seek times. In my case, hardware RAID 1+0 was necessary to meet storage performance and redundancy requirements; RAID 5/6 were simply too slow with those SmartArray 6 controllers.
You’re welcome to PM me if you’d like to discuss this further.
It would be VERY interesting to understand the rational behind this compatibility chart.
Not so surprisingly, the “cheap” drives are recommended for the “cheap” Drobos (Drobo, Drobo-S, Drobo-FS).
But such a recommendation should be based on technical compatibility, not market acceptability.
And it is VERY difficult for me to reverse engineer the rational for this table :
[list] for years, Drobo Inc. has pretended that the lack of TLER (Time Limited Error Recovery) was not an issue on Drobo “because it is not a RAID”. But they never justified such an assertion, which is hard to believe when you understand what TLER is and why its lack can be problematic when a second layer of error detection/correction is imposed on top of a disk set
 similarly, hard to understand why the Load Cycle Count (aka LCC) should be an issue on some Drobos but not all, when all of them are running similar flavors of Linux
[*] longevity, supposed to be better for “professional” drives, should not be a compatibility issue at all, since the whole purpose of the Drobo is to overcome the disastrous consequences of a disk failure[/list]
At that point, I would be sorely tempted to assume that this “compatibility chart” is more marketing bull…t, the split being :
[list] on cheap Drobos one pretends to ignore the consequences of some missing disks features (because prosumers can afford more easily to lose data ?; or are supposed to be more gullible ??), since the added cost of “professional drives” would price the Drobo out of the market
 on more expensive ones, where “cost is not an issue”, the need for such disks features finally emerges…
[*] or alternatively, there is no more need for (some ? all of ?) those disk features on high end Drobos, but recommending them gives Drobo Inc. more credibility when selling to professionals used to “the more expensive, the better”[/list]
If Drobo Inc. was more transparent about the detailed technical criteria for a “fully compatible Drobo disk”, that would also allow Drobo users to investigate utilities which may allow activation of “professional features” on lower prices disks (TLER, LCC, …) since most of those are perfectly able to support them, they are just inhibited by default by the manufacturers to force “rich professionals” to buy the more expensive ones
Revoking support for a configuration that was previously recommended and sold is a drastic move. This is not a decision any company takes lightly.
I believe that Drobo is attempting to avoid more serious costs. With this sudden change, they’re effectively trying to shift those costs to their customers. We can only speculate what those costs might be, and why they wouldn’t want us to know.
If I had a DroboPro with “not supported” drives, I’d be examining my warranty paperwork and considering whether it was worth the effort to get Drobo to help replace those drives.
Reasons I’ve seen other companies reduce their scope of support include:
Severe operational problems were identified in certain configurations. Customers were notified and offered assistance to change to a supportable configuration. (Several large tech hardware vendors)
As support was outsourced, third-party contracts fell short of original commitments to customers. Customers were notified of an upcoming change (effective at annual renewal) in support options to close that gap.
Undesirable performance caused by flaws that would be expensive to correct. (Several auto companies)
Warranty service costs that were much higher than expected. (I worked at one computer company with a high failure rate on a certain model of motherboard, including replacement parts. Techs were encouraged to replace other parts to run out the warranty period.)
Note that in most of those situations, customers were contacted and offered assistance. The auto examples I’m thinking of (Toyota I4 engine sludge, Saturn engine oil consumption) were attempts to manage costs by pushing warranty claims to the courts. The dodgy computer company was insolvent and its owners were merely running out the clock.