FreeNAS 9.2 is out
freenas.org
freenas.org
That's been my main thing holding me back from building a beefed up home server/NAS box. Friend of mine used to have an all-in-one HTPC/NAS/webserver box that would put close to $30 a month on his electricity bill. (Although you could argue he needed to run the heater less as a result..)
That's the sign I needed. Gonna take the plunge. @rasmus have you been on the 9.2 RC's and now the release or are you on 9.1? I've been having significant issues with 9.2. It's the best software out there to roll your own NAS though. Nas4Free is pretty rad too but falls significantly short when you compare it side-by-side with FreeNAS.
Also, regarding LAGG/LACP, I noticed some pretty radical throughput increases after experimenting it on my home network. It took me a while to decide on a Layer 2 managed switch, but I ultimately went with the TP-Link SG3216 (new from Amazon, pretty cheap for a 16port gigabit managed device) and it's been great.
I think that once a connection is open, it will route through one single interface, so it won't help much between two boxes. But, as you said, having multiple devices connecting and backing up simultaneously will take advantage of the multiple links.
Lowcost home file storage,
HP Proliant Microserver (mine is N60L me thinkz) Can get cash back on it if you look around (sometimes 35% back) ECC memory 8gb max (recommended if going for ZFS) Western digital reds 3tb x 4 (reds have 3 year warrently and built for raid) ZFS in raid 2 i think. (one disk redundancy)
Total space is about 7.9tb
Running FreeNas from a usb inside the case.
it works, and cost wise it was cheap minus the drives.
I was not certain about ECC being particularly important when using ZFS, so I dug [0] and discovered a multitude of intriguing resources about ZFS robustness.
Apparently, ZFS has no built-in detection or recovery mechanisms where memory corruption is involved.
HardOCP [1] led me to an academic investigative study about ZFS robustness [2]. Many conclusions were drawn, including:
In the last section we showed the robustness of ZFS to disk corruptions. Although
ZFS was not specifically designed to tolerate memory corruptions, we still would
like to know how ZFS reacts to memory corruptions, i.e., whether ZFS can detect
and recover from a single bit flip in data and metadata blocks. Our fault
injection experiments indicate that ZFS has no precautions for memory
corruptions: bad data blocks are returned to the user or written to disk,
file-system operations fail, and many times the whole system crashes.
Additional sources about ZFS and ECC can also found at http://en.wikipedia.org/wiki/ZFS#Data_Integrity.-
[0] https://www.google.com/search?q=zfs+ecc
[1] http://hardforum.com/showthread.php?t=1689724
[2] http://research.cs.wisc.edu/adsl/Publications/zfs-corruption...
So basically, you've got until newyears to get one ordered.
CCLonline[2] and ebuyer.com[3] are offering the servers for £180, so £130 after cashback.
I've got one already at home and plan on ordering two more tomorrow. Surprisingly powerful boxes, can take upto 16gb DDR3 (ecc unbuffered), internal USB and proper hardware virtualisation.
[1] http://h41225.www4.hp.com/UK_focus/PA0074%20-%20HP%20Microse...
[2] http://www.cclonline.com/product/108949/704941-421/Branded-S...
[3] http://www.ebuyer.com/430446-hp-proliant-g7-n54l-2-2ghz-micr...
PS> "zfs in raid 2" -I believe you mean raidZ, the zfs equivalent of raid5. raidZ2 (zfs raid6) would only leave you with approx 6tb usable space rather than the 9 you're quoting. As the microserver has only 4 drivebays (well, 5 if you use the optical bay) you'd probably get better speeds and same real space by using the zfs equivalent of raid10 rather than raidz2.
It will also work win _non_ ECC ram -though if you've got ZFS on the go it's highly recommended.
1. Buy the same amount of (but larger) hard-disks 2. Replace them one by one 3. After the last disk is done, the size will grow. 4. Sell your old hard-disks.
Much more painful then other raid solutions.
http://www.itsacon.net/computers/unix/expanding-a-zfs-pool/
zpool add -f poolname raidz[or raidz1,raidz2,etc] /dev/driveid /dev/driveid /dev/driveid ...
You'd then have a pool with another raidz section within it, whose size was represented by the capacity of each of the sections within.
http://alp-notes.blogspot.no/2011/09/adding-vdev-to-raidz-po...
[edit: perhaps there should be a listing of "replies added since you hit reply"-function (with a "confirm" to reply anyway-button) to cut down on redundant replies.. and trade them for upvoting reasonable replies...]
Edit: Looks like mbell and brianlweiner posted just before me. Bah :)
I'm a FreeBSD user and tend to avoid any automagic between me and the system but the FreeNAS 9.2.0 RC looked very promising.
I probably should have better documented by upgrade-in-place procedure but at the time I didn't know whether it would work. It was fairly straightforward, using documentation for root on ZFS.
This is the newest equivalent product: http://www.newegg.com/Product/Product.aspx?Item=N82E16859107...