Considering this build... Please advise :)

[Dura Ace]

Looking for low power and a small footprint. Will only be used for storage and data access ie no transcoding or anything of that nature.

U-NAS NSC-800 Server chassis
ASRock C2550D4I Mini ITX Server Motherboard DDR3 1600/1333/1066
Crucial 8GB (2 x 4GB) 240-Pin DDR3 SDRAM ECC
SeaSonic SS-350M1U 350W ATX12V

 

[Yatti420]

I'm not a fan of Mini ITX or ASRock.. Seems fine otherwise.. Keys being lots of ECC ram (can you support 16gb+?) Is the Seasonic proper form factor? Do you own a UPS?

 

[Dura Ace]

Thanks for reply. Board has four mem slots so I'll take your advise and double up on RAM. Love the case and it's driving the rest of the build (Mini ITX and PSU Form Factors). Also, liked the board as it has 12 SATA on board thus avoiding extra raid card. Yes on the UPS.

 

[beemaster]

Please go ahead with that board. It's very interesting how well it performs.
I advice you not to use 4GB memory sticks, because you will end up with just 16GB max.
The board supports up to 64GB. So go at least with 8GB sticks.

 

[Madd Martigan]

I have that board and I am using 8GB modules for a total of 16GB at the moment. I am primarily using it to serve up iSCSI targets for an ESXi host. Primarily I have a (6) 2TB drive RAIDZ set that has it's space allocated to a single iSCSI extent. I have added an additional Intel dual port NIC and I am running a single iSCSI portal with four interfaces configured in it on four separate VLANs. On the ESXi host I am using a four port Intel NIC with four separate vswitches, four separate vmkernel ports on four separate VLANs. All NICs are 1Gbps. I have Round Robin multipathing configured and I have seen up to 200MBps transfer speeds when I have enough things hitting that LUN. Obviously single VM activities are restricted to a single NIC for transfer but overall the solution works quite well.

 

[joelmusicman]

Definitely go with 8GB mem sticks regardless of how much you're using right now. Very little cost difference and they're more future-proof.