M1015 for external enclosure?

[Tiberizzle]

Does anyone use a 8087-to-8088 bracket adapter (e.g. http://www.pc-pitstop.com/sas_cables_adapters/AD8788-2.asp) with their M1015 and an external enclosure? Are there any signal strength issues resulting in excessively high CRC errors?

I'm currently using a multilane SATA bracket (4x e-SATA to SFF-8470) with e-SATA port multipliers in my external enclosure, but I'm looking to move to a SAS solution (UEK-23601 and M1015). The e-SATA setup seems to suffer from excessive CRC errors under load, which I've always attributed to connector and cable loss. There are shorter 8087-to-8087 cables available than e-SATA for connecting to the bracket and I believe SAS signaling is more robust than e-SATA, so I am hoping it's less of a problem.

Alternately, is the M1015 the wrong controller to be considering? The M1015 has internal ports, but even with the ludicrously overpriced adapter seems likely to be the most cost effective solution from my research. Is there a 2 external port non-RAID SAS controller that's reasonably priced with good expander and 6.0Gbps compatibility which I may be overlooking?

 

[dilidolo]

I use IBM 46M0907 SAS Controller which is rebranded LSI 9212 4i4e connecting to my Supermicro 936E1 chassis.
Note, flashed with LSI IT firmware.

If you need dual port, you can find equivalent to 9200-8e.

 

[Tiberizzle]

I haven't been able to find any OEM equivalents to the 9200-8e.

I was hoping to take advantage of dual linking, although it's not strictly necessary. It seems that for around $180 the M1015 can be made to provide 2 external ports (with a bracket) that support dual linking. The M5025 seems to be the only IBM model with 2 external ports at a cost of around $600. The LSI-branded 9200-8e runs around $330.

 

[dilidolo]

Unless your stoage chassis has dual expander backplane, you won't be able to use dual external port. If you have multiple chassis, you can cascade them.

 

[Tiberizzle]

There is evidence to suggest that the CK23601 supports dual-linking, in spite of Chenbro support claims to the contrary.

http://hardforum.com/showpost.php?p=1037763507&postcount=60 provides basic evidence of dual-linking functionality between M1015 (SAS2008) and CK23601 (LSISAS2x36)

http://www.lsi.com/downloads/Public/SAS Expander ICs/LSISAS2x36/SCG_LSISAS2x36_PB_043009.pdf product brief suggests that the chipset is a fully switched design capable of dual linking and even more complex topologies (PDF includes a diagram illustrating simultaneous disk multipath, controller multipath, and cascade capability)

If it doesn't work that's fine, but I'd like to test it. I have no need for internal ports, anyway, so it would be nice to know if the 8087-8088 bracket will degrade the signal excessively or if there is a cheap OEM SAS2008 external 8-port board floating around anywhere.

 

[pjkenned]

I haven't been able to find any OEM equivalents to the 9200-8e.

I was hoping to take advantage of dual linking, although it's not strictly necessary. It seems that for around $180 the M1015 can be made to provide 2 external ports (with a bracket) that support dual linking. The M5025 seems to be the only IBM model with 2 external ports at a cost of around $600. The LSI-branded 9200-8e runs around $330.

May help: LSI RAID Controller - HBA to OEM Mapping

I personally don't like adding SFF-8087 to SFF-8088 brackets unless I have to. More cables/ cable length is usually not a good thing.

 

[Tiberizzle]

Here's a dumb question.

I use Supermicro CSE-M35T-1B 5-in-3 SATA backplane modules. In the SATA Port Multiplier setup, I have 5 wires running from each port multiplier to the backplane's activity LED pins. I don't see any activity LED pins on the CK-23601 SAS expander. How would I get the 5-in-3 modules' activity LEDs to work in the SAS expander configuration?

edit: nevermind, apparently there's a drive activity pin in the SATA connector that the backplanes should work with regardless. I wonder if my existing setup would work without the activity LED wires.

double edit: yanked out the activity LED connection and the LEDs still blink. In my defense, I did preface the post with "dumb question"

 

[Tiberizzle]

I've been doing some reading and I don't think there's any reason to suspect the bracket will be an issue as long as attention is paid to the cabling.

The number of connectors in the Controller-Bracket-Expander signal path is the same as in the Controller-Backplane-Drive signal path, so there isn't really any reason to suspect the bracket as a potential source of signal degradation beyond specification.

The cable selection is important as cables tend to be designed with higher wire gauges at shorter lengths. Two cables with particularly "cost effective" gauges for their individual lengths could result in unacceptably high insertion loss over their combined length.

So, I'm going to use a 30awg 0.5m 8087-8087 to the bracket from the controller with a 28awg 1m 8088-8088 from the bracket to the expander and see if it works.

 

[Tiberizzle]

So the M1015 came in today, without a bracket , but with a 46M0864 feature key installed.

I'll be using OS softraid, so I guess I'll have to sell the feature key to recoup the ludicrous cost of the bracket from SCSI4ME.

 

[Aesma]

When I was searching for this card, someone posted a link to a bracket, it was something like 5 bucks.

 

[Tiberizzle]

So just in case anyone was curious about the CK23601, internal/external bracket, Chenbro's claims about the "Out" port, etc.:

I tested with only the expander's "Out" (i.e. cascade-only, according to Chenbro) port connected to the controller. This configuration definitely works.

I have both the "In" and "Out" ports connected now. I have no idea how to ask the controller if it's aware of the dual links or if it's using them, but I can tell you that on the expander both the "Out" and "In" external SAS port groups flash their activity indicator LEDs when the attached disks are active. So, I assume the controller is aware.

The expander and controller negotiate at 6.0 no problem, so the SFF-8087/SFF-8088 bracket seems to be in the clear. The particular bracket I linked appears to be very well designed, with short traces that are in a board layer between two ground planes. Others I looked at had much longer signal traces on the top layer.

Some of the drives are negotiating at SATA/150 and others at SATA/300, though. Strangely, it's both the drives in the #3 slot (middle of the backplane) that negotiate at 300. The rest at 150. I may try looking for higher quality fan-out cables, but it's not a huge issue with the disks I'm using now.