Drobo

Unbelievably slow drobo (Drobo,mac, osx 10.6.4, intel, fw800)

I just don’t get this. My drobo is unbelievably slow, and I need help to track down this problem. It’s more or less useless in daily use - The only thing I can use it for is nightly backups (which also run for “ever”). I also had the same problems when the drobo was connected to my previous iMac.

Here are some info:

Mac: iMac, brand new, I7, 2.93GHz, 16GB RAM Drobo with 2 1.5TB and 2 1.0TB drives. Filesystem: HFS+ Drobo Firmware: 1.3.6 Drobo Dashboard reports: Used: 1.50 TB Free: 1.65 TB It also reports everything as green and healthy
FW-section from System-profiler:

Rev2: Manufacturer: Drobo Model: 0x2 GUID: 0x1A620291822005 Maximum Speed: Up to 800 Mb/sec Connection Speed: Up to 800 Mb/sec Sub-units: Rev2 Unit: Unit Software Version: 0x10483 Unit Spec ID: 0x609E Firmware Revision: 0x10200 Product Revision Level: 2.00 Sub-units: Rev2 SBP-LUN: Capacity: 8.8 TB (8 796 093 022 208 bytes) Removable Media: Yes BSD Name: disk1 Partition Map Type: GPT (GUID Partition Table) S.M.A.R.T. status: Not Supported Volumes: Capacity: 209.7 MB (209 715 200 bytes) Writable: Yes BSD Name: disk1s1 DROBO: Capacity: 8.8 TB (8 795 749 048 320 bytes) Available: 7.14 TB (7 137 160 364 032 bytes) Writable: Yes File System: Journaled HFS+ BSD Name: disk1s2 Mount Point: /Volumes/DROBO

XBench test-results (have run these many times, and the results are pretty consistent (horrible)):

Disk Test 18.37 Sequential 16.51 Uncached Write 24.10 14.79 MB/sec [4K blocks] Uncached Write 26.50 14.99 MB/sec [256K blocks] Uncached Read 6.80 1.99 MB/sec [4K blocks] Uncached Read 63.08 31.71 MB/sec [256K blocks] Random 20.70 Uncached Write 23.07 2.44 MB/sec [4K blocks] Uncached Write 10.81 3.46 MB/sec [256K blocks] Uncached Read 37.13 0.26 MB/sec [4K blocks] Uncached Read 32.79 6.08 MB/sec [256K blocks]

I would appreciate anything pointing me in the right direction here to track down what could be wrong.

Have you tried benchmarking with USB or using a new FW800 cable?

Is it the only FW device on that port?

Have you run the tests using Kona and the ‘standard’ test set?

Thank you for taking the time to reply.

  1. I tried testing with USB now, and the numbers seems even worse:
Disk Test	13.03	
	Sequential	16.25	
		Uncached Write	21.45	13.17 MB/sec [4K blocks]
		Uncached Write	12.35	6.99 MB/sec [256K blocks]
		Uncached Read	12.22	3.58 MB/sec [4K blocks]
		Uncached Read	27.17	13.66 MB/sec [256K blocks]
	Random	10.88	
		Uncached Write	29.07	3.08 MB/sec [4K blocks]
		Uncached Write	4.11	1.32 MB/sec [256K blocks]
		Uncached Read	41.64	0.30 MB/sec [4K blocks]
		Uncached Read	15.13	2.81 MB/sec [256K blocks]

I’ve ran 3 test, and they all showed similar numbers.

  1. Yes, the drobo is the only FW800 on the (only) FW-port

  2. What is the ‘standard’ test set?

  3. I was unaware of Kona. Is this “AJA KONA System Test Version 6.0.1” on http://www.aja.com/support/konaNEW/kona-3-3x.php?

I’ve run a couple of tests, all of them while connected as USB2.

I’ve run it’s “Disk Read/Write”-test. The first time I got ok numbers (around 10MB/s, but unfortunately I forgot them), but running it again got the numbers to drop radically (to Write: 1.2 MB/s, Read: 3.9 MB/s)

I’ve run it’s “Sweep Binary Frame Sizes”, and got the following chart (the first data-points for read/write were very very low):

I’ve run it’s “Sweep File Sizes”, and got the following chart:

Thank you for taking the time to reply.

  1. I tried testing with USB now, and the numbers seems even worse:
Disk Test	13.03	
	Sequential	16.25	
		Uncached Write	21.45	13.17 MB/sec [4K blocks]
		Uncached Write	12.35	6.99 MB/sec [256K blocks]
		Uncached Read	12.22	3.58 MB/sec [4K blocks]
		Uncached Read	27.17	13.66 MB/sec [256K blocks]
	Random	10.88	
		Uncached Write	29.07	3.08 MB/sec [4K blocks]
		Uncached Write	4.11	1.32 MB/sec [256K blocks]
		Uncached Read	41.64	0.30 MB/sec [4K blocks]
		Uncached Read	15.13	2.81 MB/sec [256K blocks]

I’ve ran 3 test, and they all showed similar numbers.

  1. Yes, the drobo is the only FW800 on the (only) FW-port

  2. What is the ‘standard’ test set?

  3. I was unaware of Kona. Is this “AJA KONA System Test Version 6.0.1” on http://www.aja.com/support/konaNEW/kona-3-3x.php?

I’ve run a couple of tests, all of them while connected as USB2.

I’ve run it’s “Disk Read/Write”-test. The first time I got ok numbers (around 10MB/s, but unfortunately I forgot them), but running it again got the numbers to drop radically (to Write: 1.2 MB/s, Read: 3.9 MB/s)

I’ve run it’s “Sweep Binary Frame Sizes”, and got the following chart (the first data-points for read/write were very very low):

I’ve run it’s “Sweep File Sizes”, and got the following chart:

Try:
Read/Write
1920x1080
1gb
Disable File system cache

and check out this

OK. I’ve run several tests with the parameters above. It seems to me they really are showing there is something serious here, because they don’t look like the ones in the thread mentioned above.

test-results:

I ran a total of 10 tests, all showing charts like these.

If anyone feels their performance is slow or hampered any way please open a support case.

I just did.

Install (manually) firmware 1.3.7.

It may (or may not) improve greatly your numbers.
Before making any new measurements with AKST, be sure you let the Drobo stabilize (only way to know it is idle is to touch the drives for lack of activity, since when Drobo does its internal disks housekeeping, no light is activated).

I did and didn’t notice any improvement. I’m still testing my drives one by one to see whether one of them could be slowing down the entire array (but it takes a day before I can proceed to the next one).

Corentin

I also did without noticing any differences in performance.

@cortig & fmr : sorry 1.3.7 did not improve your performance, although our symptoms did look similar.
Either I am a special case, or something else I did between 2 AKST runs provided the improvement.
Too bad I cannot imagine what it is… (outside of an unlikely spurious relayout).

Yeah, I had hopes. Oh well, we’ll see how my tests on individual drives go. Today, disk 3 of 4. ! more to go after that.

Corentin

any findings?

So far nothing. I still have no clear explanation as to why I experience the slowdowns.

Corentin