Is it possible or advisable to use 'regular' not-sanctioned-by-the-server-manufacturer hardware in high end servers? Often these manufacturer-supplied parts have a very high price markup, and I wonder if it's always necessary (understanding that they probably apply more rigorous requirements to this hardware).
For example, Dell sells 300GB 15,000rpm serial-attached scsi drives for a certain server family for almost $600 each, while newegg sells a drive with the same specs for almost half the price http://www.newegg.com/Product/Product.aspx?Item=N82E16822116059.
Do we really need to pay these high markups, especially for disks that are likely RAID-ed and so guarded against catastrophic failure?
If you're buying "high end servers" why would you want to dilute the value of your hardware by putting in generic parts? Part of the value of buying expensive hardware is the support you get from the vendor and the extensive testing their hardware engineers perform to certify hardware. Really that's what you're paying for.
The only two reasons I know of to use "brand name" versus COTS are these:
All things being equal, I'd go with COTS for small shops, and brand-name for large installations.
Also note: the "list' price from HP/Dell/IBM is almost always reduced for corporate purchases - a "volume" discount.
You pays your money and you takes your chances...
For the most part, I buy the more expensive components from the manufacturer (in our case, that's HP). Sure, for hard drives, I could buy top quality, bare drives for less, but then I'd have to get carriers for them, mount them, and keep a few spares available since I wouldn't be able to call up and say "ship me a new one."
Basically, you're paying for: warranty (and knowing that you'll get a spare for the life of the warranty), and convenience.
Some manufacturers tend to use their own firmware on certain parts to make sure you can't add on your own drives to a server.
You usually get the same warranty on an HDD by buying off of Newegg, and better pricing for RAM and HDDs.
If you know how to build servers, you're going to want to buy all the parts yourself from a good supplier and assemble it yourself. It may end up saving you hundreds and even thousands of dollars in some cases.
If you're buying high-end servers from a supplier like HP or Dell, you will need to know a good sales person that can give you some decent discounts, as you may end up saving more than 15% on the total price from the one on the online configurators.
Some people are under the impression that Dell or HP actually make the stuff inside. This is untrue, as they only make chassis, and assemble parts from other manufacturers, sometimes re-branded parts like the PERC raid cards from Dell.
One particular manufacturer that I like is SuperMicro. They are excellent when it comes to bang for buck, and they actually make motherboards, power supplies and chassis which are very good.
The only thing you lose on this is extra money, once. What you gain by purchasing certified hardware is total support for the entire system.
Just imagine, you bought a server from Dell, and then some drives cheaply direct from WD and maybe some Kingston RAM from a guy on the corner
Then something happens, you call Dell, and they tell you that unless you remove the unsupported stuff from the server, install some supported RAM and HDDs, and then fail diagnostic tests, they cannot help you, because the stuff you have in the server was never tested to work inside it. Then you call WD, and they tell you to call Dell, and the guy who sold you the RAM is long gone as well. Other server manufacturers might actually tel you your warranty is void because you installed uncertified hardware and they are no longer responsibe for the damage it may have done. I only know for certain about Dell :)
Not all generic parts are of the same quality level.
By the time you've sourced top-quality third-party parts, the price gap may be substantially narrower than you anticipate.
Plus there are the warranty implications that others, correctly, note.
Many will not agree with me but I believe that sharing an alternative view is important here. I am speaking primarily in reference to commodity hardware. When proprietary and non x86 is in scope, things change quite a bit.
If the OEM discovers you're using your own hardware in a server, they might attempt to not support it. However, it is often easy enough to remove it for the purpose of troubleshooting. Realistically, if you are dealing with business support, usually you do not have to go through asinine troubleshooting that you would with consumer support. However, if you did not example a high technical level, you might.
You have hardware people out there that only buy hard drives on a company's spec sheet. In most cases, this level of detail is going to be excessive. This used to matter more. Depending on the hardware in question, you might encounter more fickle controllers or mainboards but typically if they share the same standard interfaces it works. I've known Intel OEM servers to be fickle occasionally but this does not happen very often.
In certain IT shops, you will have a management ethos where they attempt to isolate accountability outside of the IT department. This is often done with support contracts from large vendors. This is more difficult to do if you do not follow the vendor's support requirements to the letter.
If the staff in your IT department is less experienced with hardware, it's easy to buy what the OEM gives you and know it will work. This is a benefit for many as well.
The way I look at it is that you're paying a premium for intangible benefits. Those intangible benefits are more valuable for some than others. In a smaller shop with limited budget, I would stretch it as far as I could.
Now, if you have an existing RAID you should certainly do you best to match the drive manufacture and model. Mis-matching drives has performance implications and is generally unwise.