Incremental Build

[cmfisher4]

All,
A bit of a lead-in to start: I've been trying to digest as much as possible on these forums and get an idea for a solid home FreeNAS build. I've been convinced on ECC RAM and RaidZ2 and all of the hardware need to support this. Currently, I do not have existing hardware to build that server, but I do have hardware sufficient to build a server. Right now, I'm running a Windows 8-based "server" (hey, it was easy and cheap) and would like to incrementally build my way to FreeNAS nirvana. I got lucky without knowing it and run a i3-3220 (yea ECC-support) for the Win8 build and, based on my current and perceived future needs, plan on sticking with it. I am pretty religious about running backups (I used two separate hard drives that back up my important stuff daily (alternating) using CrashPlan) and would be willing to risk the server going down (backups covering me as much as possible) while I slowly migrate to the system I eventually want.
So, after all of that, my question is should I wait until I have enough $$ and buy everything in one shot, or can I get away with a more incremental build? I was thinking of first purchasing the Mobo (Supermicro X9SCM-F) and a single stick of solid 8GB ECC RAM (narrow this down with a bit more research and I understand dual sticks is better, but don't want to do 2x4GB and then limit myself in the long run). I would use my existing i3, hard drives (2X 2TB consumer), case, etc. I would then probably go with the second 8GB stick and the 4 NAS drives (3 or 4TB, not sure yet). My biggest concern would be the change out of hardware and screwing up FreeNAS (like you can screw up Windows when you change out major pieces of hardware).
Curious of what you guys think and I want to thank everyone for all of their input over the entire forum. Awesome wealth of knowledge and opinions.
Thanks,
Chris

 

[Nindustries]

Hello Chris, welcome to the forums!
I would only say building incrementally is a problem when there are newer/better technologies incoming you want, or when things should get a lot of cheaper.
And I guess neither is the case, unless you want a new cpu?

Me, as a student, also suffered this problem. Should I already start collecting stuff so I have my server when I get my own place? No, because the timeframe in my case is too long. It'd take too long for me to get all parts because I still study. But for you; no problem!

Now the hardware side of things.. as long as there isn't a compatibility issue between your older and newer parts, go for it.

- Sent from mobile.

 

[cmfisher4]

Nindustries: Thanks for the welcome!

I think this is the setup I'm going to ultimately end up with:
Mobo: Supermicro MDB-X9SCM-F-O
RAM: Crucial 16GB (2x8GB) ECC UDIMM (compatibility with this mobo checks at Crucial's site)
HDD: WD Red 3TB (x4 for RaidZ2, providing 5.5TB of storage...I struggle to fill up 1TB now so this should give me LOTS of room to grow)
Case: Fractal Design Define R4 (I have a FD Node 304 for my desktop and that thing is awesome...too tight for a server, IMO, however and this is $109 with $30 off and free shipping from Newegg until 10/21)
PSU: Antec 450W (already on hand)
CPU: i3-3220 (already on hand)

In keeping with my original post, I would get the mobo, RAM and case to start, using existing hardware to finish up the system. This should get my data more secure than it is (still running backups, of course) except I'll have to run a 1TB mirror (2X1TB) until I can afford for the Reds. This will require at least two iterations of disk wiping; the first to go from what I have now (Win8 Storage Spaces) and then to get to the RaidZ2 when I get all 4 of the reds. I should have enough disk space laying around to act as transfers. I have to accept the danger of losing data during transfer.
In the longer run, I may transistion to a smaller PSU to get more in the "sweet spot" of efficiency (~350W) and maybe an actual Xeon (I do intent to do streaming with this thing).
Request thoughts. I'm still running FreeNAS is a VM to get my head around it a bit more, but I think I understand enough to get the system operational and be just a little bit dangerous.
Thanks,
Chris