Drobo FS benchmark and comparison to Drobo 2nd gen FW800

just finished some benchmarking for the Drobo FS to get some more facts about the Drobo FS.


Nice work!

Just updated my post with a jumbo frames test, the Drobo FS performs even better then! Only problem is Apple screwed their Core i7 and Core i5 iMac customers in building a $2.500 machine that can’t support jumbo frames!

Hello, I ran the same benchmark settings as you with/without jumbo frames enabled. I seemed to be consistently 4mb/sec lower than each of your benchmarks, achieving 31mb/s write and 39mb/sec read. I’d be curious to see other peoples benchmarks as well so I we can try and determine if its the network factors, or our PC’s, or simply the variation in the drives/drobo FS.

I have an early 2008 MacBook Pro, 2.5ghz, 4gb ram, jumbo frames enables over GbE. Drobo FS with 3x 1.5 TB WD green drives, 31% capacity, dual redundancy disabled.

Did you enable jumbo frames on all components Macbook Pro, switch, Drobo FS?

Currently I do not have the FS connected through a switch, it is direct to the laptop. Yes I enabled jumbo frames on both devices.

Got Mac Mini 2.53/4GB connected to AAE with Cat5e. DroboFS with 3x WD20EARS gives 20mb/s both ways. Same speed when connected directly to the Mini.
Fans are noisy - Got replacement coming…

My AJA test results using specs from my sig file and AJA settings (Unix, 1024x frame size, etc) above from original article/link:

AJA DiskWhackTest - 1024.0 MB Write: 34.5 MB/s Read: 44.8 MB/s
Drobo FS - AJA Benchmarks - Jumbo direct connect

That’s Drobo FS, direct GigE to 2007 iMac (no switch/router), Jumbo Frames enabled on both Drobo FS & iMac (MTU@9000), 20% space used on Drobo FS. Edit: BTW I have to say I’m pretty pleased with that performance so far, seems to be in-line with advertised specs and article review test I read. I’ll have to see how it does once more space is used, and also if I pickup a GigE/Jumbo router, but for now IMO seems good.

Out of curiosity, I wonder if benchmark apps other than AJA would produce different results or are they (the good apps) all more or less the same?

FYI updated above with link to screenshot of benchmark stats

Quite impressive. Im no way near those numbers. Have you tested without JF through your AEBS?

Oooh not yet, I’m afraid it might look pretty bad without Jumbo, but I’ll repost if/when I give it a shot just to compare (mine all direct-connect w Jumbo so far).

FYI, looks like all Core i5/i7 Macs dropped Jumbo frame support. Same thing on a MacBook Pro 2010 i7. I’ve also discovered (the hard way) that you don’t want to try and run a mixed jumbo/standard network if you’re using Apple Airport or Time Capsule products - they tend to crash.

I’ll run some benchmarks on my 60% full FS over a network (real-world, with wireless, gigabit, switches, and the whole nine yards).

Looking forward. Youll find my tests here.

In short, about 25MB/sec write, 35MB/sec read. No jumbo frames for this test, although when I previously used them I saw roughly a 5MB/sec improvement in both read and write.

I dropped wireless from consideration - even on a 5Ghz 802.11n connection I couldn’t keep it above about 9MB/sec read/write, which is good for wireless, but not so good for testing the Drobo. Not sure if this is a limitation on the wireless or the Time Capsule’s incredibly shitty ethernet routing (I can only get 9MB/sec over a hard-wired connection to that thing).

My hardwired Mac Mini was another story. This is a 1.83Ghz Core Duo 2006-vintage Mac Mini, Cat-6 ethernet to a NetGear 16-port gigabit switch, Cat-6 to a NetGear 8-port gigabit switch, and finally Cat-6 to the DroboFS. DroboFS is populated with 2 x 2TB Western Digital Caviar Green, and 1 x 1TB Hitachi Deskstar (pulled from an older Time Capsule). So a reasonable switching fabric, good cabling, several connections to traverse, and no routers in-between. As far as I can tell none of this caused any slowdown, as my numbers compare favorably with others’ direct connection numbers.