Questions about fault indicators and SAS controller in home-built NAS

Question

For info, I'm looking at building a NAS and have explored various 4U rack-mount chassis that support 24 drives. I plan to use SATA not SAS drives and ZFS on either FreeNAS or Nexenta as the OS. As such, I do not plan on using hardware RAID. Clients would access via CIFS shares.

Obviously, I'm trying to balance cost with size and speed, in that order of importance. I've been burned going too cheap, but I can definitely not afford purchasing an integrated system.

For the chassis, there's cheap and not so cheap. These are examples only. One factor that differentiates the more expensive range is the inclusion of fault/rebuild lights on each individual drive tray, which would be very handy.

I have a few questions related to this.

1) Are the fault lights controlled by the backplane, the SAS controller, OS or something else? Also, I imagine the drives themselves have to report the error, so I'm wondering if SATA drives would support this, or if it's a SAS drive-only feature. I can't seem to figure out what keywords to search for in order to wrap my wits around this aspect.

2) There looks to be several different options for hard drive controllers, and I can't figure out what exactly I want. What I plan to do is connect the SAS controller to the backplane with SFF-8087 cable/connector, 1 cable for each 4 drives. Am I correct in assuming I want a HBA and not a RAID adapter configured as JBOD, or something else? I've seen the terms SAS Expanders and Port Multipliers thrown out there, and this quote from Wikipedia is only making things worse for me:

The correct term for the components that allows a computer to talk to a peripheral bus is host adapter or host bus adapter (HBA). On the other hand, a disk controller allows a disk to talk to the same bus. Those two are often confused, especially in the PC world.

3) Are there any other considerations it appears I am not taking into account based on the above?

What I'm thinking is I'll initially start with 16TB of storage over 8 drives and with a single 8-port SAS HBA (such as an LSI SAS 3081E-R), then add a second and third when I add an additional sets of 8 drives, the idea being keep it as fast as possible without throwing down for 15k SAS drives.

Answer 1

1) On the super-micro case (I know from experience) the fault-lights aren't connected to anything. It's just a red LED that leads to a pin-header that they expect you to attach to whatever raid-card you have. (I like the areca cards) The activity LED however IS built-into the backplane. I wouldn't expect the cheap-case to be any different... perhaps it doesn't have the activity LEDs' logic built-in.

2) I won't lie... I love the Areca cards. They're fast, reliable, and not overly-crazy-expensive. As far as the HBA/Raid discussion... I would really suggest you go with a hardware-based raid option, simply because it offloads a lot of work onto the card leaving the server's CPU to run the OS. Typically, once you leave the 2-4 port "HBA" cards... you end-up with a RAID adapter anyway. Also, if you go with Areca (I'm sure other manufacturers have something similar) they have a spiffy "multi-lane" feature which will make 2 cards work together rather than as independent controllers. (I swear I am not an Areca salesman... or any kind of salesman) Whatever you choose, make sure you get a controller-card that supports delayed spin-up. If you try to spin up 24 drives at 1 time, you'll burn up nearly any power-supply. staggered spin-up will spin up a few disks at a time and allow the jolt on the power-supply to be much lighter.

3) as for thinking about what you haven't thought of... there's no shortage of things to consider.