Generic PCs For Corporate Use? 606
porkThreeWays writes "I work for a government agency supporting about 1000 PCs. The economy has hit us just like everyone else and we are looking at ways to save money. We currently buy Dell computers and even with our government discounts end up spending about $1,000 for a pretty mediocre computer. I had the idea of building our own PCs for considerably less. We'd spec out a standard configuration that we'd use for 18 months. CPU speeds and RAM sizes may change during that time, but socket types, memory standards, hard drive interfaces standards, etc, etc would be required to stay the same. We have Dell warranties right now, but I could see just keeping spare parts on the shelf and building that into the cost of the PC. We'd also be able to transfer Windows licenses because the Dell installs are non-transferable. However, I couldn't find anyone on the large scale doing this. Is anyone on Slashdot using PCs they built themselves on the large scale?"
transferring Window license? (Score:3, Interesting)
be careful transferring windows licenses... they're all OEM licenses and the T&Cs don't allow you to transfer them to another machine (ever). Of course this is based upon my knowledge from a few years ago when i worked in the licensing field, so things might have changed (IANAL)
Re:$1000 a PC? (Score:1, Interesting)
Virtual Machines (Score:5, Interesting)
Use server based VMs or terminal servers. Then use winterms for the desktops. You can get those for a couple a hundred dollars and they last forever.
Go to Walmart or Best Buy (Score:2, Interesting)
software? dell wants like $150-$300 for office + (Score:2, Interesting)
software? dell wants like $150-$300 for office + over priced ram (Dell warranties may not like you having 3rd part ram)
the school I work for used to... (Score:1, Interesting)
Mgmt killed the white boxes and bought IBM (now Lenovo) for half the machine at twice the cost. Failure rates are roughly comparable (lenovos probably have a slightly higher failure rate).
So, whatever was involved in that decision making process, cost of the hardware and reliablilty wasn't paramount. I think it is because my workplace is such a political environment, and they perceived risk to themselves by buying no-name.
There is no cost savings (Score:3, Interesting)
Unless you are some intern making $5 / hour or something, the amount of time you will spend assembling these things will far outstrip the cost savings.
IE - say you save $200 / machine. How many hours will it take you to build that? Three? Four? Now figure in how much you make per hour. Your "savings" are out the window.
Re:$1000 a PC? (Score:5, Interesting)
I did this recently. Not with 1000 machines, mind you, but five. Dell wanted an exorbitant amount for the machines, insisted that since we were getting hex-core processors that we must get discrete graphics, and a bunch of other technologies* that we just didn't need.
By going with Newegg and building it myself over a weekend, the price was cut in a little more than half.
*We do scientific number crunching, but don't have any GPGPU code right now. Our codes fit in an average amount of memory, are CPU intensive, and take up very little hard-drive space. Dell couldn't understand selling us a hex-core CPU with a 80GB hard drive. Further, we couldn't specify the number of PCIE slots (in case we do GPGPU later on), but they did insist on discrete graphics, which we absolutely didn't need. This quote came from their SOHO line. On the true "server" side of things, their prices are astronomical.
Re:$1000 a PC? (Score:3, Interesting)
I raised this proposition to my managers as well, and got laughed at. Well, not really, but it wasn't taken with very much consideration at all. You have to remember that these are people who have been trained to believe in the free-market economy and capitalism and that it is their duty to conduct their business to support other businesses, otherwise the market as a whole is harmed. This, of course, is until there is a profit to be had from not supporting another business. They are loathe to cut out a middle man unless it means a substantial guaranteed return on investment. Building your own workstations doesn't guarantee a return on investment. At least not in the reality of office-land it doesn't. Maybe a small shop could do it, but it actually has more potential to lower the ROI.
What are you commenters talking about? (Score:1, Interesting)
If you build your own computers, you are still getting roughly the same parts that Dell/HP would slap in there for a considerable amount less. How is the failure rate going to be higher? It's not! It will be equal. The difference being that you need a tech guy who can repair them when they do fail versus having to call in a Dell/HP tech.
Instead of telling this guy "Oh, this is a bad idea, I tried it and it didn't work." Give him the reasons why. I'm willing to bet the guys that are saying this work for Dell or HP and are just trying to scare him. After all, every sale counts, right guys? If you build a computer right, it won't fail.
And what does a government agency need 4gb of ram and a PCI-e video card for? Check their emails? Search a server side database? Manipulate a spreadsheet? Give me a break. The computer that can get all of these jobs done can be built for as low as $275 each. Pick up an enterprise copy of Windows/Office and be done with it.
Building Your Own in a Corporate Setting (Score:5, Interesting)
Re:Hardware vs Total (Score:3, Interesting)
At the company I work at, we get a lot of Grants & Contracts. Those grants and contracts will pay for all sorts of labor with no problem, but any equipment we have to pay for out of our own profit margin. Therefor, we tend to focus on the reverse: Put people to work and pay as little as we can for equipment. This means that if we can save any money in expenses, even at a cost of labor... as long as that extra labor cost isn't extreme, then we pay for the labor getting the equipment for cheap.
We get some very powerful machines for about $500 in equipment costs. We probably spend another hundred or two in assembly and support labor. But the assembly and support labor is money we'd be spending anyway.
use AMD chips (Score:2, Interesting)
my reply (Score:2, Interesting)
If I read this right, my suggestion would be to look at frye's
http://www.frys.com/template/computerspc [frys.com]
From then (ownership on) you can use the old dell hardware for spare parts or add them into the new pc (memory, CD /DVD/ NIC / VID CARD, POWER SUPPLY, etc. when it can be used.
Do it. (Score:1, Interesting)
I work in the higher education field, and we've been building computers for the last five or six years. We spend about $600 per computer (that was the last build), for a decent, just over the tip of the average computer. We add in about a 10% margin, and keep those parts for repairs.
I suggest that you look at what Dell, HP, Gateway, and the other pre-built PC companies are selling, aim for the medium range, and spec-out what it would cost to build the exact same thing yourself. Remember to add in the cost of time/labor, otherwise you'll throw off your figures when you present to the boss.
Re:$1000 a PC? (Score:5, Interesting)
Dell has better volume discounts than you ever will, both with Microsoft and the hardware manufacturers. They further offset this by bundling in a whole load of crapware on the default OS install.
Even after accounting for their profit margin and your time spent re-imaging the machines with a clean version of Windows, the cost from Dell compared to DIY for standard beige-box business machines should be somewhere between slightly cheaper and slightly more expensive; if it's the latter, a single point of contact for warranty issues is still perhaps worth the money. If it's the former, you win on all counts.
If you are a large customer, Dell sends you a machine, you do a clean install, create a standard image with just the software you want on your machines, send the image to Dell, and they put it on all the machines you order from them. No bloat, no time wasted customizing each machine, and no extra charge for the service. Its especially nice if you have multiple locations and want to have a standard configuration used across the board.
Re:Don't do it (Score:1, Interesting)
I agree, but for different reasons. White box machines mean you are taking on the integration piece all on your own - you need to test every piece and all of the drivers yourself. Something broke? You figure out what it is and source another one. The benefit to companies like Lenovo and HP (Dell is crap for the most part) is they do the integration for you and may even post enhanced software and drivers that are designed to work better with the hardware they have delivered to you. In addition, management software that comes with products from Lenovo and HP make managing large numbers of machines easier. This is the true benefit in products from these companies - it's the Value Add they offer that makes the products worth a few bucks more.
Yeah, I'm doing it (Score:5, Interesting)
Our product is, technically, little more than a computer in a fancy box. We started off wanting to buy a small computer and shove it in, but found that we needed way more power than anything available at the correct form-factor.
Being computer guys, we figured we'd just build it ourselves.
Truth is, around here (Toronto), OEM computer suppliers are everywhere. Good ones (Infonec) with reasonable inventory and reasonable access, Poor ones (Tiger direct) with huge inventory and no access, and remote ones with infinite access and no inventory. So we're covered from every angle be it some rare component or an immediate same-day requirement.
Do components break? Sure. Some hardware is defective out of the box. That goes onto the reject shelf. Some break when we drop it. That goes into the garbage. Some break after they are installed when it's just not stable and it takes many hours to figure which part is at fault. Those are annoying, but they go onto the reject shelf just the same.
The reject shelf gets turned over by mail with a few RMA phone calls every few months. The nice part is that if you wait long enough, you tend to get newer models from the manufacturer, so it's winds up almost being worth-while.
The garbage is, honestly, an easy thing to avoid. Wear cotton, ground yourself, and never put a motherboard onto a chair unless you atcually want someone to sit on it.
The nice thing about 1'000 is that while you can't get much of a discount on the components themselves, you do get more than priority service from the suppliers. And that can really be valuable when it means that your deployment schedule is uninterupted.
Yes you can save money. You should wind up saving about 40% over a dell machine. Of course, you'll lose the warranty service. And that's where the trick comes in. You get to balance something that you've never balanced before.
You get to say: "cheaper = more servicing = more expensive" while also saying "higher quality = less servicing = still expensive"
Here's the trick: "higher quality = longer life-span = re-use"
The real savings aren't on those 40%, because you have to service them instead of dell servicing them. dell's more efficient (money wise) than you are. But because of that, dell's cost-optimizing the quality, because they don't get to keep it. They'd rather take the risk that the parts won't break, and fix the 20% that do.
That doesn't work for you.
You want to spend more, only saving 20%, then you want to do minor upgrades at the right now, so really only wind up saving 10%, then you want the machines to last twice as long, and be able to salvage the parts for future machines -- repeatedly. This also has service replacements of broken parts and diagnostic repair fed for free.
In the end, you wind up spending the same 100% out of the gate, you spend only 80% the second generation, and then you spend closer to 40% by the third generation.
In the end, you have high-quality machines, top-quality parts, and very few break. Service calls are not only at a minimum, but you're just swapping out the possibly bad parts with known-good parts, then checking the possible bad parts at a later, more convenient date.
You're also providing the new guy with a better computer to get him started on the right foot, you're giving the guy with a lot of work to do this week that extra gig of ram to make it easier.
But yes, this presumes that you are comfortable running such a service. It's definitely easy to do, but it's complicated as hell to keep it organized.
Re:$1000 a PC? (Score:4, Interesting)
Something else you have to consider when putting a computer together is that he's talking about the government.....they have rules and stipulations about pretty much everything. Are the parts you selected approved? Are there labor types of contracts (i.e. unions) that would get you in trouble for building or moving or plugging in the systems? Could you even get approval for the purchase?
Even if it's an option for a small business, it's a complication for the government.
Re:$1000 a PC? (Score:3, Interesting)
When they're less than $200. [slashdot.org]
Besides, I think he was offering his recent experience with building Generic vs Dell which is what the author is trying to do, not necessarily saying he must use 6 core processors.
Re:transferring Window license? (Score:3, Interesting)
Re:$1000 a PC? (Score:2, Interesting)
insisted that since we were getting hex-core processors that we must get discrete graphics,
Did it even cross your mind that perhaps the hex core boards .. did not have onboard graphics ??
was one of the "other" things you didnt need gigabt Ethernet ?? Thats like common now ..
Re:Go directly to FoxConn (Score:3, Interesting)
We've actually had some pretty good experiences with their Netboxes. They're small enough to mount on the back of a monitor and, at $450, fully equipped, they're cheap. My only complaint is that, when you actually do mount them on a monitor, the power switch is difficult to reach. If only there was some way to have the machine turn on, automatically, when you turn on the monitor...
Re:Why not start a company? (Score:5, Interesting)
Building machines to sAve money ceased making sense about 4 years ago, like SpaghettiPattern, I build systems by choice, and fully understanding it likely cost more than a comprable Dell box. People that attempt to rationalize building PCs these days don't talk about savings, they talk about choosing the co potent to get a certain benefit (better PS, upgraded video card, better chassis). For basic office work, a business class PC should last five or more years without any failures. They would likely last longer, but there comes a point where a failure is almost certain to happen, and in most cases you want to avoid the unscheduled down-time, loss of the system.
PCs are a cutthroat business, and if you think you can do a better job than Dell's robots, fine, but I doubt you're right.
Re:$1000 a PC? (Score:5, Interesting)
Well in our case (and I know it doesn't apply to the original poster), we use Linux, so that saved us even more, because getting Dell to drop Windows on a non-server build is like pulling teeth.
As for the time and pay, I got some overtime pay and got to play with some neat machines. It takes only 30 mins or so to actually build each box and I have our Linux as an image on a USB stick, so it takes another half an hour to copy the image over. All said and done, I probably only spent four hours putting the whole thing together.
I got lucky in that there were no defective parts, although one hard drive went bad about six months later. Other than that though, it's been 18 months since I did this and we haven't had any problems.
Of course, I wouldn't want to scale this process to 1000 machines, but with a little planning and foresight, it's definitely do-able in the tens of machines range.
Re:Don't do it (Score:3, Interesting)
correction: chargeout rate for 200 weekends? (Score:2, Interesting)
Whats your chargeout rate for weekend work?
Minor correction: what's your chargeout rate for 200 consecutive weekends (you did 5 computers in a weekend, the guy needs 1000 computers done...) ? Still having fun? Wife and kids fed up yet? Boss getting angry because 4 years is too slow a turn round time for the 1000 machines to be done?
Could shorten that time by hiring in 10 guys and doing it in 20 weekends but now the company needs to set up systems for managing 10 people, health and safety, insurance, admin for payment etc.
Not scaleable methinks....
Re:correction: chargeout rate for 200 weekends? (Score:2, Interesting)
The summery says they manage 1000 computers not that all the computers they manage need to be replaced. Most companies such as my own have a maximum age for PCs in there facility. We try to go with about a 4-5 year replacement cycle. If we followed this method strictly it would mean every year we would replace about 30-40 PC's. I would thus guess they only need to 200 or so PC's a year. at 1000 computers there IT staff should have at least 3 people who could squeeze in assembling machines during normal work hours as squeezing things in is what the job requires. It would just add about 33 hours work for assembling and with a proper lan imaging installer hardly any time for that portion.
The real catch here is that a $500-1000 PC from dell that normally includes the support contract at least for businesses tends to run close to the cost of building your own. Once you get into the 2k-4k range non servers you tend to see build it yourself machines being more cost effective.
Re:$1000 a PC? (Score:1, Interesting)
Rule of thumb where I work is about 1PC in 50 will have an issue of some kind
That's awesome. When I used to work at the factory of one of the top three OEMs, the goal was a 2% fallout, which would work out to one in fifty.
It was impossible to get production to try to do any better than that. 2% was the goal, trying to do better was a waste of money. :/
You could do this, and maybe save money.. (Score:4, Interesting)
..but I wouldn't want to.
Years ago, I worked as the service manager in a high-volume white-box PC shop. I was in charge of the guys who built boxes, the guys who troubleshot them, the guys who supplied inventory, and the guys who had to ship the parts back to the wholesaler for RMA.
Building 1000 boxes is six man-months worth of work, minimum. You need to figure in a 2% failure rate on finished machines, so you need to order parts for at least 1020 finished machines; you should also figure in a 3% failure rate during the build, so order enough parts for 1051 boxes. When the builds are done, RMA the bad parts and keep everything around for in-house spares.
Don't buy white-box based on warranty, because the warranty is useless after 90 days or so. You'll probably have to send the part back, it will get swapped with one that's been "fixed" and sent back to you. Half the time, that means you'll get one that somebody else returned and wholesaler's tech can't replicate the problem with, so he sends it to you, hoping you'll be okay.
As for building 1000 boxes at once, the way to do it is in partnership with a wholesaler. You'll need to rent some real estate, about 3000 square feet. Nothing dusty, and no carpets. You can probably get it cheap for 30-45 days, look for stuff that's been for sale/lease for a while. Hire 8 guys for a month who "like computers". Have the wholesaler ship you a tractor-trailer full of parts. Get a bunch of locked cabinets, lock all the RAM, CPUs, harddrives up. Stack the motherboards in a locked room. Stack the cases in a corner. Your 8 guys can unload a 40' trailer, count the parts, and put them away in ~12-14 hours.
Every morning, each guy who shows up gets his parts for the day. A low-output worker can build five boxes. A high-output worker can build 10. No drills allowed unless they have clutches. At the end of the day, they can demonstrate a working machine running windows, and start your burn-in suite. The next morning, for every box that passes burn-in, you give them $25. Then they pack the machine back up in the box the case came in, stack it in the room slated for deliveries.
Parts can be swapped 1:1, don't allow floating parts on the floor or they will never get put in a machine. Also, only allow clear garbage bags on the floor. No food or drink, either.
Oh - the reason for so much real estate? The most productive way to build machines is to use about 4 feet of table each, and to do them all at once. So, if you're building eight boxes that day, you need 32 feet of table. And 8 mice, 8 keyboards, and 8 monitors. Don't unbox mice/monitors/keyboards for your build, it will cost you time and not increase your reliability.
Make sure you provision these 1000 identical machines with removable HDD trays. That way, when one fails, you rip the drive out, stick it in a spare, and send that machine to be either fixed or pitched. Fixing might be expensive, though. Remember your assembly crew? They're long gone, and probably not very good technicians.