Clients awful server has him headed for disaster... any recommendations?

thecomputerguy

Well-Known Member
Reaction score
1,437
I have a client I picked up a few years ago after he had just purchased a new server. I warned him at the time that the "Server" wasn't really a server and was entirely inadequate to be called a server.

Specs:
Dell T110
1 x 500GB Boot Drive
1 x 500GB Data Drive
Xeon Processor
8GB RAM
Server 2008 R2
*No mirroring what so ever

I explained my feelings towards the server and he said, well this is what we have now and I don't have money for a new one so we have to make this work. My estimation is that this server cost him just over $1000.

Today he calls me because his C drive is full. I login and find out the Boot drive had been partitioned into a 90GB Boot partition and a 380GB or so partition of empty space. Easy enough fix I just expanded the volume and made the C drive the full 500GB's.

While I was in I took a look and basically told him that IF the server crashed, because there is no mirroring, whether it be the data drive or the boot drive BEST CASE SCENARIO he's looking at half a day of downtime with worst case being a full day until we could get the server operational again. Obviously the server will not continue to be operational if there is no active redundancy in place.

The server does not reboot cleanly most of the time and he has a Carbonite for business mirror going to an external, but carbonite crashes all the time (weekly), and it requires a reboot to get carbonite up again, and who knows how good this mirror actually is anyways.

I told him just like I told him 3 years ago, the server is inadequate to be called a server, and it is just a glorified workstation, and he needed to spend at least a few thousand to get a real server in place.

He declined again and said, OK well I can't afford downtime so is there anything you can do to make this server more reliable. I told him I'd ask around but I really don't think there is a way to get this server any more reliable than it already is.

I thought about dropping in a RAID card and setting up a RAID1/RAID1 - BOOT/DATA setup but I'd probably have to reinstall Server 2008 and setup everything from scratch and that just sounds like a nightmare especially considering the database applications he runs.

Then I thought about doing the lamest setup possible, and dropping in 2 more 500GB drives (I don't even know if the chassis will handle 2 more drives), and using the built in Windows software mirroring to get the RAID1/RAID1 going, but then I just slapped myself realizing how dumb that is.

Any ideas here or do I just tell him he's doomed and he needs to spend some serious money or just continue to operate as is knowing that he's probably headed for disaster at some point.
 
He doesn't need to spend "serious money".... if by serious money you mean several thousand dollars.

He can get a really nice used server off of eBay, a nice dell poweredge or a hp proliant machine that
will be leaps and bounds better then what he has for a few hundred dollars. Spend a few hundred more
dollars on new, proper hard drives for in it and you should be in great shape. It's used yeah, but it shouldn't
be much more then $1000 in hardware. That could probably carry him a few years, and just let him know that
he should plan to move towards better hardware as he can afford.
 
He doesn't need to spend "serious money".... if by serious money you mean several thousand dollars.

He can get a really nice used server off of eBay, a nice dell poweredge or a hp proliant machine that
will be leaps and bounds better then what he has for a few hundred dollars. Spend a few hundred more
dollars on new, proper hard drives for in it and you should be in great shape. It's used yeah, but it shouldn't
be much more then $1000 in hardware. That could probably carry him a few years, and just let him know that
he should plan to move towards better hardware as he can afford.

Yes serious money for him is several thousand.

I don't do the used thing. If the server I provide isn't warrantied I don't put my ass on the line for it even if it's just a recommendation.

I do appreciate your response though, just not my cup of tea.
 
If the existing OS is already experiencing crashes and problems with booting those would be the first things to address.

Performance would be the 2nd. That machine could be expanded to 16GB of RAM for example.

Then if the owner still doesn't want to invest in a new server sell him 4 new hard drives:

1. Make sure the Dell Perc S100 drivers are already installed in his Server OS.
2. Do a "bare metal recovery backup" to a USB hard drive, make sure you get the Data volume.
3. Install 4 new decent 1TB drives. WD Purple Surveillance drives? $61 each at Newegg.
4. Use the embedded Perc S100 controller to create 2 new mirrored volumes.
5. Perform a "bare metal recovery" back to the newly created mirrored volumes.
6. After booting use Windows Disk Manager to extend the volumes to reclaim the new space.

This is the kind of project I'd do on a Saturday after ensuring the Carbonite backup ran successfully the night before. Even if things don't work out you can always slap the original 2 drives back in the machine while you contemplate your next move.

I wouldn't recommend trying to use the existing old drives as part of the new mirror. Best if the drives match right down to the firmware revision.

As mentioned by the previous poster he could have a fine used server for not much more money although the Server license could be an issue.
 
If the existing OS is already experiencing crashes and problems with booting those would be the first things to address.

Performance would be the 2nd. That machine could be expanded to 16GB of RAM for example.

Then if the owner still doesn't want to invest in a new server sell him 4 new hard drives:

1. Make sure the Dell Perc S100 drivers are already installed in his Server OS.
2. Do a "bare metal recovery backup" to a USB hard drive, make sure you get the Data volume.
3. Install 4 new decent 1TB drives. WD Purple Surveillance drives? $61 each at Newegg.
4. Use the embedded Perc S100 controller to create 2 new mirrored volumes.
5. Perform a "bare metal recovery" back to the newly created mirrored volumes.
6. After booting use Windows Disk Manager to extend the volumes to reclaim the new space.

This is the kind of project I'd do on a Saturday after ensuring the Carbonite backup ran successfully the night before. Even if things don't work out you can always slap the original 2 drives back in the machine while you contemplate your next move.

I wouldn't recommend trying to use the existing old drives as part of the new mirror. Best if the drives match right down to the firmware revision.

As mentioned by the previous poster he could have a fine used server for not much more money although the Server license could be an issue.

All of the server even the POS T110's come with a built in PERCS100 chipset in the mobo? Didn't know that. ...
 
Just curious what did you use to re partition the 90 gb drive to add more room?

Just Windows Disk Manager?
 
With multi core CPUs....there is nothing wrong with Windows Server software RAID. I would take that...over the built in fake-RAID setups out there (which..that Perc S100 likely is). Windows Server does RAID 1 quite well. yeah...years ago, with single core CPUs....it was quite a performance hit. But not anymore with multi core CPUs. I would take built in Windows Server software RAID over any fake-RAID controller that does not have its own dedicated processor and BBU (battery backup cache). A controller without those 2 things is NOT a RAID controller..it is fake-RAID. And I've seen waaaaaaaaaay more headaches from fake RAID than with Windows Server software RAID.

I've boosted "glorified desktop" servers performance by swapping their drives out with WD Black drives or RE drives. Those are enterprise class SATA drives, a full steady 7,200rpm and importantly 64 megs of cache, and 5 year warranty. For this guy here on a budget I'd just do Blacks...since the server is already over 3 years old...as the RE drives are a bit $$$ier.

Clone existing volumes to 2 new drives...put the 4 new drives in, built your RAID mirror within Windows Server disk manglement.

How many users?
What database?
 
Just curious what did you use to re partition the 90 gb drive to add more room?
Just Windows Disk Manager?

Windows will move partitions around by itself now (finally). No need for other utilities if you don't want although it's a knee-jerk reaction to reach for GParted.
 
I think we need to know how many users and what's on the server to properly advise. Saying the thing needs to be replaced because it isn't a real server... well, what does he NEED a real server for? As an IT consultant you need to be cognizant of computer performance, reliability, and value to the client. A 5U rack server with full redundancy is probably not good value to this client.
 
We ended up deciding on 4 x WD RE 1TB Drives in a RAID1/RAID1 configuration using Windows RAID ... thanks stonecat!

The server holds basic docs, pdfs, etc. and additionally 2-3 custom programs using access type databases with a total of about 4 users.

@YeOldeStonecat
 
For a couple of these budget servers we have replaced the drives with WD Velociraptor. Due to their 10K spindle speed the performance increase is quite significant.

WD class them as workstation drives rather than enterprise... but they come with a 5-year warranty and the same 10^16 URE rate (or 1.2m hours MTBF) as the RE drives. They also come with enterprise class features such as TLER so the work in RAID setups.

Any reason these aren't commonly suggested? I think they are the perfect drives for budget servers.
 
Any reason these aren't commonly suggested? I think they are the perfect drives for budget servers.

I love 'em. Back when they first came out..I used the 36 gig one in my home gaming rig. And then the 74 gig ones.
My current office workstation has the 2.5" 300 gig model...in the "ice cooler" housing.

But re: Servers....you raise a good question. Guess my response is.."size"...to my knowledge, the 300 gig is the biggest one?
Granted...could work for many smaller clients...but we're just used to having more space/size I guess.
 
Ask him how much it will cost him per hour of down time then compare how much time it will be down with your preventative measures with the amount of time it will be down if it crashes...and then throw in possible data recovery costs.

Personally, to reduce down time, I'd go with setting up a new server and migrate the data over.
 
What @lcoughey said. How much lost revenue and ****** off customers will it cost?

As far as used machines. Some machines on eBay come with warranties. Including OEM's. The first one I bought still had 1.5 years left. And some food for thought. If he can't come up with the cash for the server the major OEM's are always happy to finance.
 
We picked up a used Dell (R710?) recently with remaining warranty, and a mighty good thing - part of provisioning is updating all BIOS, firmware, etc. which included updates for the firmware in the power supplies (!?). That part did not go smoothly, bricked the power supplies. Dell sent replacements under warranty, but in the meantime we had to borrow (with permission) a power supply from another customer who had the same hardware.

Heck of a nice machine for the price. Preloaded with 72GB, dual 6-core Xeons, etc. Only disappointment was PERC6i not H700, and I don't think it has the Enterprise management.
 
Last edited:
But re: Servers....you raise a good question. Guess my response is.."size"...to my knowledge, the 300 gig is the biggest one?
Granted...could work for many smaller clients...but we're just used to having more space/size I guess.

The latest generation drives come in 250GB, 500GB or 1TB.

I can pick up the 500GB model for just over £90 but the 1TB model is over £200... quite a jump.
 
Honestly, I would have gone a step further in this situation than just setting up the software RAID (though I agree, it's the cheapest and simplest way to get him to at least hard drive redundancy and isn't as bad as it was in the 2003 era). Our company almost always slaps in VMware ESXi first on any server we deploy if they are unwilling to spend money on vSphere.

In this situation, I would have recommended performing the bare-metal backup suggested previously (though I would prefer a real backup tool to the one Windows has -- too many limitations like it can't handle 2TB+ disks). Then I'd add either 2x 1TB drives or 2x 500GB drives depending on his willingness to spend just a little extra for more space down the road. Next, create 2x RAID1's with the resulting drives if using 1TB drives, or RAID6 with all 4 if going with 500GB drives. Next, install ESXi on the box (completely free if using free license). Finally, set up new VM and recover the old server to the VM.

With this setup, not only do you get the hard drive redundancy you wanted, but you also get more options with the VMware setup. You can make more adjustments to the virtual hardware, move the VM to better hardware in the event he has the money to do so, and get a second VM on the existing hardware if necessary (though likely would need more RAM for that).

Using VMware ESXi on servers with cheap clients has given me so many more options down the road than installing Windows directly on the box. I only ever do anymore when a software vendor tells me they won't support it, which almost never happens.
 
Next, create 2x RAID1's with the resulting drives if using 1TB drives, or RAID6 with all 4 if going with 500GB drives. .

For servers....having a pair of drives for the hyper-v host provides far superior performance. A pair of RAID 1s...or for larger setups, I do a RAID 1 and a RAID 10...or larger ones with many guests..I'll do a pair of RAID 10s.

Put servers system volumes on the first RAID, and then the data volumes on the second RAID. Pagefile.sys across both. This way it's multiple spindles...and the concurrent read/write is far superior over having multiple server partitions on a single spindle (such as a single RAID 5 or 6).
 
Back
Top