Reasonable budget recommendations

[Thefinaleofseem]

I'm looking to build a decent FreeNAS box, but budget is a bit of a constraining factor. I'd like to keep costs down where possible without severely compromising performance or reliability. A bit of background on what I'd like to do...

I undertook a project a while back to rip every movie I have with MakeMKV (DVD and BluRay) and tossed them onto a drive. This worked out well, but I started hitting limitations, especially since these are not recompressed (I've seen BluRays in the 40GB+ range). I'm working on compressing them to a more manageable size, but this will likely be for local storage on mobile devices and, of course, is going to take quite a bit of time. Even then, I'll be keeping the larger files for the local HTPC. The primary goal of this NAS is to be a home media server for some pretty beefy videos. However, I'd also like to do some general backup on it (certain pieces of data from my own computer that I'd like to have kept in multiple places, probably ISOs of all software I have on disc, etc). I'm not looking to use a Plex Media Server, just basic storage. If at all possible, I'd like this NAS to saturate a gigabit connection. If it could do that in both directions, even better, but that's not a big deal. Reads are definitely going to be more common than writes, so if a tradeoff of slower writes with faster reads is necessary, that's okay. I should also stress that I'm not planning on putting any mission-critical data on here. This is a home NAS that will likely not contain data that isn't restorable from other sources, although granted, it would be a pain in the butt to do so. Reliability and redundancy are great, but I don't need the ultimate system here.

So I'd like to toss a NAS together with some ZFS goodness, although I don't need excess bells and whistles. I'm aware that ZFS without ECC is like trying to squeeze out a fart after some bad Taco Bell (sure, you may be okay for a while, but soon enough you're going to be in deep crap). However, buying brand new Xeons, a Supermicro board, etc is just not in the cards at this point. The upfront cost is too high. I'd like to hold out until I can reach that point, but, well, I've already thrown together a FreeNAS box as a short-term solution, and I'm not even going to go into the details of what's in it beyond that it's old, low-end, and UFS-only with a hodgepodge of drives in JBOD. Basically, it was what I was able to cobble together with what I had at the time. I'd like to ditch that near-term solution sooner rather than later.

So I'm looking to trim costs without seriously compromising the NAS. Low power consumption is great, too, and I'd like to stick to that if possible. Low noise is a must, so I can't just snag a surplussed rackmount. As such, I'm looking to eBay for hope, and I may have found some. I know that some cheeks might clench at that notion, but if it can be done...

I'm looking at the following, albeit not set in stone, of course:

Xeon L5630 - Low power, not too spendy. 1366 seems to be about as good as I'll get without a steep cost increase. The one concern is clock speed vs more cores and which would be better for this system. If a faster dual-core is going to be the better choice, then I can do some more digging.
Supermicro board, possibly 8XSTE - I know that a motherboard on eBay is potentially a crap shoot. I intend to test it thoroughly before using it and buying only from sellers that have a long, solid history of moving this sort of hardware. Single-socket 1366 should be dandy. I did see that Intel has some low-end CPUs that support ECC, but finding a solid board/RAM combination with that seems a bit sketchy, and it still means $200+ on a motherboard.

As for the rest, there's really no skimping. RAM is going to be whatever I can find on the compatibility list for any given board, hopefully no more than $150 for 3x4GB, which leaves three more open slots if I need them. A decent case with enough drive bays, 80-Plus Gold PSU, a few 4TB WD Red drives, etc. Those costs are relatively fixed no matter how I cut it beyond looking for good sale prices. However, it seems that cash can be saved by going a bit older on the CPU/board without going too far back that I'm compromising power consumption and performance. I considered some later-model Core 2-based Xeons, but it seems that they aren't nearly as good at low-power idling as Nehalem or later, not to mention DDR2 FB-DIMMs and the rest of the system.

I'm just wondering what folks might think of this and cores vs clock speed in this particular case. Again, this is a home NAS, not a production server for critical business data. I know that it would be better in the long run to go with new and modern components, and I wish that were feasible at this point, but it aint so. I'll probably buy everything but the drives initially and use my existing drives before picking up the Reds later.

 

[Scareh]

well before cyberjock comes inhere telling you what to do, let me put in my 2 cents.
Basicly, you get what you pay for. If you go for lower end hardware, you're going to end up with lower end performance. (i learned that the hard way).

some extra info you're going to have to provide:
1) how many 4Tb drives are we talking about? 5-20? (this determines the amount of ram you "should" ideally get)
2) how are you going to primarly acecss your nas? Cifs/ftp? (keep in mind cifs are singlethreaded)

In short, spending that little more amount of money would prevent the deep shit you're talking about. Also, don't forget an UPS (i know its not your most valuable data but still, must be a pain to rerip everything :/)

 

[cyberjock]

I won't come in here.. I read this yesterday and I decided I wasn't going to touch this...

 

[Thefinaleofseem]

I'd be using perhaps four 4TB drives in a RAIDZ. It'll be a long time before I need more than 12TB. Nothing major. I would think that Nehalem would do well enough for this, but my experience with Freenas is limited to the aforementioned Q&D box I'm using now, which I really want to replace with something more proper sooner rather than later. I realize that this scenario isn't ideal (I don't really blame Cyberjock as I figured this may look bad to some of the savvier folk), but I'm trying to make do with what I can scrounge up.

 

[cyberjock]

Just so you understand, my problem isn't with your choices. The stuff you've discussed seems realistic and safe. The problem is that people that have a post like yours generally also does a bunch of stuff that isn't safe, but they didn't realize it at the time and didn't mention it because they were 100% sure it was safe. Later, when they lose their data and ask what went wrong and we tell them they are shocked to hear how lacking their knowledge was on the topic. Unfortunately, that makes a mess because:

1. You probably aren't about to discuss in excruciating detail what you plan to do.
2. I don't want to have to read a 10 page report on what you plan to do.(that kind of time is reserved for my consultation service)
3. I don't want to give the impression that I endorse something that may be unsafe.

Generally, when people start talking about price and say things like "the upfront cost is too high" there's no easy 'cheat' to get to the end-game. If there was we'd be talking about that. ZFS isn't for everyone. If those upfront costs are too much for you then consider going to an alternative. But all too frequently people thinking they are going to cut costs and still be safe find they didn't cut as much as they had hoped, it wasn't safe, or it wasn't even compatible.

Personally, I think it's better to walk away knowing you don't want to swim in the shark infested water than to think you're going to put on your pajama's, jump in the water, and then be upset when you find out that pajama's don't protect you from the sharks.

Good luck.