Anyway, I am having even more problems with the TR4M not coming up when the system boots. My method of powering up the eSATA units 10 seconds before the PC has stopped working for the TR4M. So far I can’t get it to work today (I only see the orange power LED, no Green host connect light once the bus scan detects the HBA). Since I have too much work lately to do I don’t have time to mess with it anymore. Going to replace it, and use the TR4M on another PC that rarely gets rebooted (security camera system controller and CIFS backups for the other 4 systems I have. I will send an email to tech@sansdigital to see if they have an answer (or even do answer).

It has been long enough that I do not remember some details: I had to return the drives, but they let me keep the enclosure. I had at one point 2 drives and later 3 drives in the unit. iirc, I had drives installed opposite (top down?) of what one responder indicated they needed to be. Worked fine.

Currently, it is not in use as I lost 3 drives in one day. 2 were in the Sans Digital and 1 in my computer. As a result, I have nothing ATM to test with.

I do not use RAID arrays much anymore, I’m more a JBOD user now, since is independent of having to match drive models. So I can not discuss much on that. In my review, it worked quite well, but it was of course 4 drives w/the provided controller.

When you say the Sans Digital works only 3132 controllers, I get the impression that this is RAID setups? If I am wrong there, correct me. Assuming that, what about JBOD/San Digital/non-3132 controllers? Go or no go?

I bought some extra hardware at the local FRYs yesterday to try an experiment. I added an internal 1TB Seagate drive and backed up everything on my 2 TowerRAID volumes to it to start with. I also bought an IOGear GICe702S3R5 2-port PCIe HBA which is also an SiI 3132 chip just like the rebranded SiI HBA included with the TR4M. I have 3 eSATA attached DAS units to experiment with: TR4M with 4 500GB HDDs, RAID-0 is one; the Termaltake MUSE R-Duo with 2 400GB HDDs, RAID-0 is the second; and a Vantec eSATA enclosure with 1 320GB HDD is the third (these run too hot after a few hours of moderate use for a work disk on my system).

I did have an SIIG PCI HBA with 2 eSATA ports (not an SiI 3132 ASIC) already in my main PC before I added the HBA from the TR4M. I had the MUSE and Vantec attached to that original HBA. Once I added the TR4M to the system, port-1 on the SIIG stopped working, and port-2 was eratic. Some sort of system conflict that I couldn’t find. Hence the removal of the SIIG HBA (it was just PCI anyway), and the installation of the new IOgear PCIe HBA ($30).

All 4 eSATA ports work reliably now, although I still must power up the TR4M unit about 5 seconds before the PC power, or the TR4M will not get picked up on the port scan (no green light, no disks visible).

As an experiment, I moved the TR4M to a port on the IOgear, and moved the other units over to the TR4M’s HBA. All units worked fine. I was wondering where the config info was kept for the RAID configurations, and evidently it is on the multi-port logic in the TR4M unit itself. I would have to install another IOgear HBA in a second PC here to verify that this is not captured in the Registry somewhere, but I doubt that is the case.

BTW: When I backed up the TR4M file systems to the new 1TB Seagate I timed the operations. The first volume on the TR4M contained 80.2GB of data (82,370 files, 6353 folders) and averaged 51MB/s (since this was a sequential operation). The second volume had a lot more very large Zip files (2-3 GB each), so the sequential efficiency was higher. That volume had 184.3GB (107,834 files, 10,887 folders) and it averaged 65MB/s. Hence, read performance out of the TR4M (in RAID-0)is very good, as is the write performance to the 1TB Seagate.

I haven’t had time yet to run multi-threaded IOmeter tests and conduct small block random tests (an IOPS measurement) to compare the 1TB Seagate with the TR4M and the MUSE boxes. As random (IOPS with a Response Time) and sequential (throughput as MB/s) workloads are at the opposite ends of the spectrum, a unit-unit-test can behave very differently. A sequential workload is the easiest test for any system controller, but it also shows up the back-end design limitations (larger arrays) as well.

BTW: At FRYs yesterday, there were 5 returned TR4M units on the shelf, and the sales dude who manages that section said those people couldn’t get more than one disk to work. So they were also tripped up on the nature of the box and required use of the 3132 controller.

Thanks for the clarification on the SI chipset restrictions to use this device. My Express Card adapter has a 3132 chip, and the last comment suggests it should work. Going for it!

I will point out one thing. No where in my review do I indicate anything about hardware Raid. Everything at the beginning refers to software-based Raid. I would hope that most users/readers would understand that one is not going to get hardware Raid at this price point.

And FWIW, I did plug this directly from the original eSATA card to another one w/o a problem. Granted, it was another variation of a 3132 chip. From comments here and from Sans Digital, it also works with the HP MSS units. I would be curious to know what eSATA chip they use.

Disregard the comments about the chart sizes. Just noticed that they pop up when you click on them – duh.

Say, I can’t make out the scales on the charts. I would recommend either saving these figures as higher res embedded images, or make them larger, or create a simple summary table of results.

In my backup tests (WinXP SP3, Explorer copies) from internal disks to the TR4M (in R-0 mode, 4 500gb7200RPM disks), I found that it could absorb about 38 MB/s on sustained writes (about 360GB and > 120k files and folders). Reads were about 45MB/s.

In R5 3+1 mode, I found that writes were about 12MB/s, and reads about 23 (same large copy workloads). R5 costs 4 disk IOs per host write, so no surprise there, except that the SI HBA doesn’t perform full stripe writes with heavy incoming workloads. But, for the money, probably shouldn’t expect that.

I would ask you to change the comment about the TR4M’s HBA – it is the guts of the TR4M and not a freebie eSATA interface for those who don’t already have one.

Thanks for putting together the review. It was helpful. Maybe add a comment to point folks to the much better documentation at the SiliconImage site.

No problem. Actually, I have no ego in this – I barely care except for the teaching aspect to help others avoid problems – what I do most of the time. I stopped being ego-centric way back when I passed 50. You apparently react with hostility when your view of the world is challenged (a classic sign of imprinting – not internalized knowledge). You threw up a scattered handful of what you thought were facts, but then didn’t like being challenged on your accumulted ideas – that is an ego-centric at play. So why are you so defensive when you encounter someone who has experience that is far above yours? That is a truism in everyday life. There are those who know more and have done more than I, and I am glad to chat with them since one can always learn new bits and pieces along the way.

I noticed you had no comment about the TR4M reboot issue, something useful for folks who might read this review.

BTW: Since you mention it, the CX3-40 is a good entry-level midrange box for an IT dept for edge solutions and such.