Main Area and Open Discussion > Living Room
Building a home server. Please help, DC!
Stoic Joker:
Servers need to be rock stable, and that requires well matched components. They don't need to be constantly tweaked to maintain lightning speed...They ain't for that.
DIY desktop can be fun...DIY server = bad idea. The factory takes great care to make sure all components are matched and play well together. And they're also very well documented and easy to get parts for.
Googling an error message for a mainstream server will get you tons of spot-on info.
Googling an error message for a DIY server, will get you tons of cryptic guessing games.
superboyac:
Servers need to be rock stable, and that requires well matched components. They don't need to be constantly tweaked to maintain lightning speed...They ain't for that.
DIY desktop can be fun...DIY server = bad idea. The factory takes great care to make sure all components are matched and play well together. And they're also very well documented and easy to get parts for.
Googling an error message for a mainstream server will get you tons of spot-on info.
Googling an error message for a DIY server, will get you tons of cryptic guessing games.
-Stoic Joker (May 24, 2011, 12:41 PM)
--- End quote ---
Thanks. I needed to hear that. That's the impression i got the last couple of weeks while I was looking into it.
40hz:
I'll agree again with Stoic's comments. (Jeez! It's getting to be a habit lately... :mrgreen:)
From an assembly perspective, a server is no more difficult to build than a desktop.
But it is hard to get a good match of components unless you really know what you're doing (as in: you're familiar with a lot of server-type brands and products) or you're willing to spend some time reading and digesting Scott Mueller's massive Upgrading and Repairing PCs book every time you run into something about hardware you don't fully understand. Server hardware is similar enough to desktop hardware that you can make bad decisions if you only think of a server as some sort of 'souped up' PC. Because the entire design criteria is different than that of a desktop.
Then there's the infamous HCL (i.e hardware compatibility list) Microsoft publishes for server system builders that you'll need to consult if you're going with a Windows solution.
In short, more trouble than it's worth unless you have something very specific in mind. But unless you have an exotic application, web, or database server requirement (like Google or some ISP hosts do), it's always easier (and usually cheaper) to buy an assembled system. You could go bare-bones and get something from Super Micro. But for the same or less money, you could also get a very capable used server (with warranty) from a 'name brand' maker.
Kind of a coin toss...
I've pretty much stuck to the big three prior to using Dell: HP, Compaq, IBM.
Of the three, I liked Compaq (ProLiant series) best, then HP, then IBM.
Today, I'd stick almost exclusively with Dell or HP - with my current preference being for Dell.
8) :Thmbsup:
--------------
P.S. if I were engineering a truly massive data center (which I'm not) for something like Facebook or Amazon, I'd definitely call IBM and invite them to buy me lunch (which they would) to discuss their "heavy iron."
;D
Stoic Joker:
You could go bare-bones and get something from Super Micro. But
-40hz (May 24, 2011, 06:51 PM)
--- End quote ---
lol My website is hosted on a 5 year old Dual Xeon SuperMicro box.
Building a home server. Please help, DC!
It been a great machine, but today I'd definitely go with Dell.
40hz:
OK...
Now that we're all finished talking common sense, let's start getting a little crazy once again in the grand tradition of the Donation Coder forum.
I mentioned earlier I didn't feel it made sense to custom build a server unless you had a specific goal or need in mind.
Here's a blog post by Backblaze (an online storage service provider) who did have a unique need for a very high capacity, high density storage server that was reliable and extremely inexpensive to build. What they came up with was not only unique - they released the full design specs and a detailed parts list so others could build their own. It uses off-the-shelf components except for a custom designed rack case. It packs forty-five(!) 1.5Gb TB hard drives for a total of 67.5 gigaterabytes of total drive space. And it costs about $8000 in unit quantities including their custom designed case.
Here's the little monster, minus its 'skin,' with five drives installed - and another 40 to go! 8)
While it would be patently insane to exactly duplicate this box for personal use, it would be relatively easy to go with a subset since the bulk of the cost is tied up in the hard drives. If you take the drives and the custom rack case out of the bill of materials (and remove some of the parts needed to handle all 45 drives) the cost drops below $1500. Pretty amazing.
Here's a video discussing the company's business model and it's server design. Interesting to watch both from a business and technical perspective.
A blog post with a cost analysis and full details on the design and construction of this server (with parts list) can be found here.
Here's a couple of illustrations taken from the blog to whet your appetite. Click to enlarge.
Building a home server. Please help, DC! Building a home server. Please help, DC!
A couple of points:
Hardware is only part of the equation when building a server. Backblaze has done the hardware spec and component matching. Which is a major time saver for any who wish to follow in their footsteps. But note that this system is also designed to run a customized version of Linux and in-house software specifically set up for Backblaze's own requirements. And they are not providing copies.
Building a cloud includes not only deploying a large quantity of hardware, but, critically, deploying software to manage it. At Backblaze we have developed software that de-duplicates and chops data into blocks; encrypts and transfers it for backup; reassembles, decrypts, re-duplicates, and packages the data for recovery; and monitors and manages the entire cloud storage system. This process is proprietary technology that we have developed over the years.
You may have your own system for this process and incorporate the Backblaze Storage Pod design, or you may simply seek inexpensive storage that won’t be deployed as part of a cloud. In either case, you’re free to use the storage pod design above. If you do, we would appreciate credit at Backblaze and welcome any insights, though this isn’t a requirement. Please note that because we’re not selling the design or the storage pods themselves, we provide no support nor warranties.
--- End quote ---
If you were planning a Windows Server deployment, you'd still want to check Microsoft's HCL to be sure there were no known problems with components. I'm guessing some of these components (like the port multiplier backplanes and a few of the cards) will not be found on the 'official' hardware list. Which is not to say they won't work. It's just you may have driver or other issues if you use them. The only way to be sure would be to buy them and test them thoroughly before you commit your data if they weren't on the HCL.
But anyway, there you have it: 68GbTB of storage platform for only $8K.
Like I said earlier - pretty amazing. 8)
Navigation
[0] Message Index
[#] Next page
[*] Previous page
Go to full version