$750 3-year warranty on server?

Let'sgoflying!

Touchdown! Greaser!
Joined
Feb 23, 2005
Messages
20,322
Location
west Texas
Display Name

Display name:
Dave Taylor
Buy or no? It is a 24/7 onsite, no mention of response time so far.
A day or two down will not cripple or kill our business.
Server is only $1800 (plus three disks which are a separate $830 but not sure if they are covered).
I truly think if either hardware of software fails, I will use my tech support to diagnose and fix (I have a call in to them to see if they can do this). Do servers have an expensive achilles heel component in them which is prone to failure?
 
Buy or no? It is a 24/7 onsite, no mention of response time so far.
A day or two down will not cripple or kill our business.
Server is only $1800 (plus three disks which are a separate $830 but not sure if they are covered).
I truly think if either hardware of software fails, I will use my tech support to diagnose and fix (I have a call in to them to see if they can do this). Do servers have an expensive achilles heel component in them which is prone to failure?
Personally I wouldn't. But I will have dual power supplies installed and an extra drive on the shelf for when my RAID 5 loses a drive. You do not want a degraded RAID array because a second drive failure will cause you to lose all the data and you'll be doing a bare metal restore. All the better if it's a hot swap spare already sitting in a bay. Those are the achilles heal IME.

I have also heard the "day or two down won't kill us" which is often only true until it happens. We don't have a redundant internet connection and sure enough, last Wednesday our ISP had a card go out in the local CO just around the corner. We were down for 6 hours. Now the CFO says "we need a redundant internet connection." Funny how the purse strings loosen when an event occurs; the appetite for down time suddenly diminishes.

Of course I know nothing of your business, my firm is much more 24/7.
 
Extended warranty? How can I lose?

What makes you think a warranty is going to have any affect on downtime or recovery?
 
What makes you think a warranty is going to have any affect on downtime or recovery?

When I mentioned downtime Ron, it was immediately after my comment about "no mention of response time".... so you can see I was not combining your two thoughts of warranty and downtime above - I was discussing 'response time and downtime'. (Perhaps it would have been better if that sentence were in brackets, following the first sentence.)
In summary, I agree with your statement.
Let's get back to my questions.
 
Thank you Greg. We have actually had many 2-5 day events of no power whatsoever, so I am quite confident we can survive a few days. Large stack of blank paper, pencils so we can do it!
I agree with your comments on having a spare disk ready to go and two power supplies. What else? Besides the automatic backups (one onsite, one offsite).
 
Buy or no? It is a 24/7 onsite, no mention of response time so far.
A day or two down will not cripple or kill our business.
Server is only $1800 (plus three disks which are a separate $830 but not sure if they are covered).
I truly think if either hardware of software fails, I will use my tech support to diagnose and fix (I have a call in to them to see if they can do this). Do servers have an expensive achilles heel component in them which is prone to failure?

It depends on the quality of the hardware and how the server is used and maintained. Good-quality servers that are properly maintained and protected, and that are not used as workstations, are usually pretty trouble-free. Bearing in mind that I've been out of that end of the business for a while, I'll offer up some more observations.

I always researched all the parts of a server before committing to purchase it. I wanted to know the exact makes and models of all the components. They all have reliability histories that can be uncovered with some research.

As for processors, server-grade Intel and AMD processors tended to be pretty reliable across the board, but I wanted to make sure the specific processor was up to the intended job and well-cooled. Also, I had fewer problems overall with Intel than AMD processors, but I only had a handful of problems with either over the years.

Motherboards, on the other hand, vary greatly in their reliability. When possible, I always felt more comfortable with a year-old motherboard model that had an impeccable reputation than a brand-new one with no track record. All else being equal, I also preferred Intel boards for Intel processors, mainly because they were rock-solid stable and I never had a problem with one. I also found Gigabyte and Asus boards built on Intel chipsets to be very reliable and stable overall.

I also insisted on good cooling of the processor, chipset, and case. If there was no chipset cooler, I would often install one. I also tended to go a bit overboard on case cooling, and part of my routine service calls was cleaning the schmutz out of the machines (especially in dusty environments) using a static-free vac, compressed air, and brushes if needed. Don't underestimate the problems that schmutz can cause.

By the way, cleaning a computer is not as easy as one might think. It's probably best left to a computer tech. Any air blowing over components could cause ESD, and rapid cooling caused by compressed air used to dislodge stubborn gunk can damage components. It's a more delicate procedure than it appears and requires discretion and ESD protection. I did the job with the machines powered down and cooled to room temperature, which usually meant graveyard shift calls.

RAID controllers, whether on-board or add-on, are a particularly finicky class of components. They accounted for the majority of data-loss incidents on RAID arrays because they tend to corrupt the whole array when they fail. That's one reason why a good backup that is not part of the array nor controlled by the same controller is essential. In the old days we used tape drives. Nowadays, imaging to an external hard drive or online backup are more common solutions.

Hard drives, of course, also should be of high quality with good reputations. I tended to avoid bleeding-edge drives that pushed the envelope on capacity. As with motherboards, I liked some good history and preferred a slightly-older drive with a solid reputation for reliability over a bleeding-edge drive with no track record. If it was to be part of a RAID array, I purchased a few identical spares, as well. I also liked drives with generous cache buffers. This was both for performance reasons and because the cache reduced wear and tear on the hard drive. (This is also important on SSDs, by the way. Even though there's no disk, they still have limited writes.)

As for the warranty itself, it's one of those things. If you use it once, it will most likely pay for itself. Twice, and it's a virtual certainty. So it's a gamble, just like any other service plan. The most important thing is the reputation of the company issuing the warranty. If they suck, so does the warranty.

If you're happy with your local tech support company and they've been in business long enough for you to believe that they'll still be around three years from now, you may want to approach them about whether they'll issue a warranty on the machine. When I was in the business, sometimes I would and sometimes I wouldn't. It depended on the quality of the machine, whether my company would be routinely maintaining it, and my overall experience with the client.

Most of the times when I issued a warranty, I made money. A few times I got burned. It's a gamble for the provider, as well; and how well the provider will respond to a call they're losing money on depends on the provider's business practices and their ability to easily absorb a loss. If your local tech support company is hanging on by a thread financially, that could be a problem if an expensive failure occurs.

Rich
 
We are looking at this HP server, I have no idea of its components.
SMART BUY ML350T09 E5-2609V3LFF SVR
 
Buy or no? It is a 24/7 onsite, no mention of response time so far.
A day or two down will not cripple or kill our business.
Server is only $1800 (plus three disks which are a separate $830 but not sure if they are covered).
I truly think if either hardware of software fails, I will use my tech support to diagnose and fix (I have a call in to them to see if they can do this). Do servers have an expensive achilles heel component in them which is prone to failure?

What is the standard warranty? IME, parts failures are typically infant mortality, or they go more than 3 years unless they are in rough service. Even the beatings and heat that our servers see either behind the bridge or in the 'doghouse' of boats do pretty well. Typically all we carry is a spare drive or two for the RAID. F it has a one year warranty that FEDEX Red the parts out, I probably wouldn't buy the extended warranty. Not even sure if a motherboard or processor costs that much, and if you have a tech, I'm not seeing the impetus to spend 40% of the purchase on an extended warranty.
 
We are looking at this HP server, I have no idea of its components.
SMART BUY ML350T09 E5-2609V3LFF SVR

I can't find much other than the processor is a Xeon E5-2609 V3, which is on the low end of the Xeon line (in terms of capabilities, not quality). But it may do just fine depending on what it is your server will be doing. Most server duties aren't especially CPU intensive. Others are.

It also says that the motherboard uses an Intel C610 chipset, which seems to have a good reputation. It doesn't say who made the motherboard or anything else about it. I've always had good experiences with HP hardware in general and servers in particular and don't recall having any problems with their OE server mobos, but my experience is not recent.

It's also not clear whether or not it has dual power supplies, which is something I failed to mention earlier.

You do know that that server doesn't include hard drives or an OS, right?

I don't have any way to really check anything beyond published information anymore. When I was in the business and frequently bought / specified hardware, I had people I could call who could find out the answers to specific hardware questions. Nowadays, not so much. Someone more current would be a better source of information.

Rich
 
Thanks Rich.
Yes we are purchasing OS and disks separately (and licenses).
 
I don't believe I have ever had a motherboard or a CPU from an HP/Compaq fail. Drives and daughter (riser) cards, yes. Used to be I only bought HP so you should be happy with your server.

My first and only outright CPU failure was on a DELL. Once the new CPU was installed those 3 DELL servers have run flawlessly for 3 years hosting a VMWare environment. Find out if your RAID controller has battery backed cache. If it does, you'll want to keep an eye on it periodically because that battery will fail.

Also, if you don't need a ton of disk space, ask if RAID 1+0 (RAID 10) is an available option. It's supposedly safer than RAID 5, but at the cost of some additional disk space.
 
Back
Top