Solved

File Server rebuild - need help with drive configuration

Posted on 2010-09-05
33
633 Views
Last Modified: 2012-06-21
I'm looking to upgrade the server hardware in our office (<15 people).  Everything works, but I want to be proactive since we're approaching the limits of our current SCSI drive capacity.  The last time I messed with it, it was a simple 30G Gateway server with OS & data on a pair of mirrored(?) SCSI drives.  The mobo got fried in a flood several years ago, but the two SCSI drives survived and I built a new box around them.  If I rebuilt this correctly, the current drive configuration is:

Server 2003 OS on SATA (II?) drive
Data (critical project info & quickbooks data) on two 30GB SCSI drives (hopefully mirrored)
Archived data on second SATA drive

This server is a domain controller and also allows employees access to approximately 26-28G of critical project information, CAD files, and bookkeeping info.  I'm not an IT guy, but it seemed reasonable to keep the frequently read/written data on robust SCSI drives, and mirror them to reduce the risk of data loss (although it's completely possible that if I didn't configure this properly they are not currently mirrored/striped/whatever.  For what it's worth, there is a systematic backup schedule of normal & incremental backups that maintains 6 weeks data on a NAS device in a different part of our building).  When I last worked on the box, we were only dealing with 12-16G of data so the 30G drives were more than adequate.  Now that we're approaching their full capacity, I want to upgrade, with the express goal of reliability and decent speeds for multiple users simultaneously accessing files (excel, pdf, bookkeeping, CAD, scheduling databases, etc).  Would the following configuration work?  My terminology sucks, so please bear with me:

Drive 1: OS, on same SATA drive or upgraded to SSD
Drive 2,3: Critical data, on mirrored (raided?) SSD drives (60 or 90G each) & PCI controller card
Drive 4: Archived Data, will reuse current SATA drive

If I’m correct, the SSD drives will reduce lags for the users, especially in CAD and accounting software (already have gigabit cards, so network throughput should not be an issue).  Is this necessary for the OS drive, or just for the actual data storage?  Am I even on the right track?

Thanks in advance.

JD
0
Comment
Question by:jdjintx
  • 11
  • 7
  • 6
  • +3
33 Comments
 
LVL 47

Accepted Solution

by:
dlethe earned 250 total points
ID: 33607534
You need so little data that SSD is a nice play because it will ultimately cost you less for an industrial-class, high reliability SSD than a pair of enterprise class disks and an appropriate RAID controller.   Goto www.ssd-reviews.com and pick something with highest reliability and best error ratings.  Don't be so much concerned with speed, as you'll never hit the ceiling unless you feel the need to run some benchmarks.

Make darned sure whatever SSD you choose has been tested/qualified with your motherboard.  Some combinations just don't work, others are quirky, so google the make/model of SSD at same time as your motherboard and see if you get any hits with people that are complaining or reporting wonderful experiences.  No need to take any chances.   Since SSD won't have the performance issues as the other drives, mirror a pair of SSDs.   Win2K3 software-based RAID1 works quite well with SSDs. No reason for a RAID controller.  In fact, unless you get a more expensive  RAID controller, then it will not only be a bottleneck, it is one more thing that can have compatibility problems, and be a point of failure.

As for archived data, personally, I strongly advise against a single SATA drive for archiving.   Just do it right and go RAID1, or if this is something you will rarely need, go with an online backup service and let it be their headache to maintain the data.
0
 
LVL 95

Expert Comment

by:Lee W, MVP
ID: 33607705
> ... with the express goal of reliability ...
As you should be for the server that runs your business... but then WHY would you BUILD the server.  Did you buy two of every part so you have spares?  

Servers run the business and should be respected - go buy a real server from a name brand (Dell, HP, IBM) and get an appropriate warranty (3 year with 4 hour response 7x24).  To me, it's very unwise running a home built server when the price difference is really minimal and having the server fail for any length of time could cost you FAR more in lost sales and productivity than the few hundred dollars you might save building it yourself.

0
 
LVL 47

Expert Comment

by:dlethe
ID: 33607778
I agree with leew strongly on this .. I just did not jump on the soapbox and kept OT.  One thing to add, however, For a few hundred bucks you can get yourself an older quality HP/Dell/IBM server that is 32-bit only, but no problem for you.  It will also have ECC memory and is designed for 24x7, and will most likely continue to run long after MSFT removes support for 32-bit operating systems.

 Mount SSD internally or behind a drive bay and just fish the connectors through (the 2.5" or 1.8" SSD will give you plenty of space to work with in the larger 3.5" SCSI drive bays)

No need to buy new server hardware, get new storage.
0
 
LVL 95

Expert Comment

by:Lee W, MVP
ID: 33607797
Sorry dlethe, I treat most questions as a request for advice, not black and white.  Call it a soapbox if you like.

I wouldn't be buying a used/older server - a warranty is just as important as the system being designed to operate 24x7.  And warranties with the appropriate response time will be even more expensive on an older machine (I never buy a 4 year warranty as the price difference is often so great that it doesn't make sense to keep the machine longer.

The only time you can ignore the warranty to some extent, in my opinion, is when you have spares.  The warranty is what gets the system up and running NO LATER than the next morning.  Or having spares that you can swap out.
0
 
LVL 47

Expert Comment

by:dlethe
ID: 33607833
Sorry leew, I was not implying it was a bad thing, I am saying I didn't jump on my soopbox this time, as you know I usually do.   Clean/tested used vs Clean/untested new is philosophical.  With a sample size of one, either system has same probability of working flawlessly, being a lemon, or being dead on arrival.   The only guarantee is that eventually it will die, so minimize the disruption by backing up and having a plan for a bare metal restore.  Used gives you opportunity (and obligation) to pre-stage a completely redundant replacement system.  New means you are at mercy of warranty support.

0
 
LVL 11

Expert Comment

by:gmbaxter
ID: 33607856
Judging by the little number of staff in the office, I doubt that there is budget for a highly redundant solution.

I would suggest obtaining an idea of the budget available for this server overhaul, highlighting the importance of the server and also throw in the peace of mind with support available on new hardware.

Once you have an idea of your budget, post back here and we can suggest some ideas, re-using what we can from your current setup.
0
 

Author Comment

by:jdjintx
ID: 33608875
All,

Thanks for your interest!  Some additional information:

1.  I understand your concerns about reliability & a home-built server not being a match, but...  all we use the file server for is domain control & data storage, and as critical as these are, it's very simple and easy to keep up & running.  I have six weeks of backups on a NAS drive, so  2-3 hours after an issue and we're up & running with the data restored on any of our networked desktop PC's.  So in answer to the question about spares- every PC in our office can function as a short-term spare.

2.  The budget is not as big a concern as the fact that I'm a DIY type that sees no need to put in a system that's overkill for such a basic fileserver.  My original server was purchased off-the-shelf from Gateway, and it came with an OEM Server 2003 license.  By cloning the drive (after the flood wrecked everything else), I salvaged the license on a SATA drive & continued to use the mirrored 30G SCSI drives to keep only the data.  This has worked very well, and I'm guessing that replacing the SCSI drives with larger SSD drives will take care of my capacity issues and improve access times for various files.  I don't know if the drive that has the OS impacts speed as much as the drives that have the data, but if it will help I can upgrade that drive as well.

3.  The archive drive is not a backup; it contains old project data that is infrequently (if ever) accessed, and is read only.  It has been backed up, so it is completely redundant and available on the server only as a convenience.  It may be used to research old projects for a customer, but is seldom used and written to only when I "retire" files form the active server to this drive.

Splitting these drives - OS, Data, Archive seems to be a good way to compartmentalize various segments of the server, but I'm very interested in hearing better approaches or technical reasons why this setup wouldn't work.  DLTHE mentioned mirroring the SSD drives without a controller, which sounds great.  Can this be done in addition to standalone drives for the OS & archived files, or do all the drives need to be part of the raid/stripe/mirror configuration?

Thanks again!

jdj
0
 
LVL 95

Expert Comment

by:Lee W, MVP
ID: 33608885
2. "... and it came with an OEM Server 2003 license.  By cloning the drive (after the flood wrecked everything else) ..."
Actually, you violated licensing and subjected your company to a huge fine.  The OEM license is tied to the hardware and dies with the hardware.  If you want the ability to transfer licenses, you should get a Volume License.  It's not much more expensive than an OEM license and it will not die with the hardware - it dies with the company.
0
 

Author Comment

by:jdjintx
ID: 33608903
Ouch.  I assumed that I was OK since I wasn't using more than I'd paid for, even though I had upgraded most of the hardware during the repair.  I'll need to look into the volume license, but I think the rest of my questions will still apply.  Is it worth looking at Server 2008?
0
 
LVL 46

Expert Comment

by:noxcho
ID: 33609422
Is it worth looking at Server 2008? - Yes. Windows 2008 R2 would be better as MS stopped support of Windows 2K family support.
0
 
LVL 55

Expert Comment

by:andyalder
ID: 33611120
You do not need a volume license, a normal retail license is transferrable, it's only OEM licenses that are tied to the original hardware.
0
 
LVL 95

Expert Comment

by:Lee W, MVP
ID: 33611277
Yes, a normal retail license is transferable, but typically more expensive.  Further, a volume license provides easy downgrade rights and the ability to download the media direct from Microsoft.  And while a retail license can be resold, it's rare that anyone buys them - difficult to trust they are legit.  
You might also want to read this:
http://www.microsoft.com/business/en-us/resources/technology/business-software/the-benefits-of-volume-licensing-discounts-and-more.aspx#Thebenefitsofvolumelicensingdiscountsandmore

BTW, I'm very much in favor of using partitions on a server.
0
 
LVL 47

Expert Comment

by:dlethe
ID: 33611761
Another twist ... first, get the right kind of license, but use your favorite virtualization software and install the O/S onto a virtual machine.  Then you are free of dealing with hardware pretty much forever, and the virtual machine can be migrated to bigger, faster machines as necessary as the license is for the virtual machine, not the hardware. (Just you certainly want to get a computer on the HCL list so you know that you won't run into compatibility problems).  Also a virtual environment makes it easy to backup, migrate, and do disaster recovery clones and such.
0
 

Author Comment

by:jdjintx
ID: 33612134
All the licensing points are helpful, and for all I know I have a retail license- I had assumed it was OEM because it was pre-installed, but I also had to increase the quantity of users from five to ten and I'm doubting that this was done with a straight OEM license.  The main problem is that it's been so long since I bought the machine from Gateway that I can't find CD's or paperwork, so cloning the drive was the simplest way to move forward.

Now that we've covered the license, what are your opinions on the drive configuration?  Did the additional information I posted help, or do I still need to answer a few more questions?

Leew - are you suggesting that I should partition one set of mirrored drives rather than installing the four drives as I described in my previous post?
0
 
LVL 95

Expert Comment

by:Lee W, MVP
ID: 33613452
Not at all - in fact, You need to analyze how much time the major functions of the server will spend doing things and then divvy them up on the drives.  For example, If I had two sets of mirrors (I assume that's what you're talking about), I would probably divide them up like this:
C, Exchange DB, Other DBs, Volume Shadow copy of disk 2 partitions, Pagefile.
D, User Data, Group Data, Exchange Logs, Other DBs Logs, Volume Shadow Copy of disk 1 partitions, Pagefile

But I would also chop up the drives - drive 1 - C: (system, boot, programs), D: (databases), E: (Exchange), F: (VSS and other Misc. Data)
drive 2 - U: (User Data), G: (group Data), L: (exchange and other DBs logs), V: (VSS and other misc data)

Of course, if you're not using Exchange, you can scratch that partition, etc.

By the way, I can build servers to.  I like a good night's sleep, so I don't build them.  It's cheaper and more assuring to go with name brands with appropriate warranties than to cross my fingers that my system stays running.  And who do you call when you have a weird problem with the RAID controller your RAID keeps going offline?  Do you call the RAID controller maker?  The motherboard maker?  The hard drive maker?  And what happens when you call them and they blame the other guy (I've had this very situation happen!)
0
 

Author Comment

by:jdjintx
ID: 33615072
"By the way, I can build servers to.  I like a good night's sleep, so I don't build them.  It's cheaper and more assuring to go with name brands with appropriate warranties than to cross my fingers that my system stays running."

I agree, but I'm looking at this as a drive replacement rather than a server build.  Would you scrap a perfectly good server rather than increasing drive capacity? I bought the original from Gateway rather than building it, and the rebuild (after the flood) was intended as a temporary stop-gap to keep the office running until a new server arrived.  The rebuild worked so well, however (and has since 2006), that I never bothered with the replacement.  In fact, the only reason I'm messing with it now is that the SCSI drives that hold the data (but not the OS) are nearing their capacity.  As I said earlier, I'm definitely not an IT guy, so maybe I'm oversimplifying the significance of changing drives in a server; is this the equivalent of a rebuild, and does it introduce significant risk?  If so, I am more than willing to buy a new server to keep things running smoothly.  Otherwise, I see several options:

1.  Keep the current setup to the greatest extent possible, which could mean limiting changes to swapping the two 30G SCSI drives (Cheetah?) with two similar drives that simply have higher capacity.  It would be necessary to ensure drive compatibility with the current PCI controller (LSI?).

2.  As above, but also replace the controller if there is any reason for doing so- perhaps drive compatibility, although I suspect MOBO compatibility is a more likely concern, and that has already been proven.

3.  Replace the data drives with mirrored SSD's that use Server 2003 drive configuration instead of a controller (as suggested by dlthe),  The OS would remain unaffected, as it is resides on a SATA drive that works just fine at the present time.

4.  As in 3, but also upgrade the OS drive to SSD if this would likely improve performance.

I may not have made this clear before, but I am pleased with how the server runs right now (with the drive configuration described in my original post), and would be perfectly happy to simply increase the size of the SCSI drives and be done.  Much has changed since 2006, however, so if an upgrade to SSD's could improve performance without compromising the current system stability, why pass up the opportunity?

Thanks again.

jdj
0
Why You Should Analyze Threat Actor TTPs

After years of analyzing threat actor behavior, it’s become clear that at any given time there are specific tactics, techniques, and procedures (TTPs) that are particularly prevalent. By analyzing and understanding these TTPs, you can dramatically enhance your security program.

 
LVL 46

Expert Comment

by:noxcho
ID: 33615464
1.  Keep the current setup to the greatest extent possible, which could mean limiting changes to swapping the two 30G SCSI drives (Cheetah?) with two similar drives that simply have higher capacity.  It would be necessary to ensure drive compatibility with the current PCI controller (LSI?).
- This is the simplest way you can go with. Image the existing configuration. Replace the HDDs with bigger ones and restore to new drives resizing the partitions on drives proportionally so to allocate space existing partitions.
And yes, you need to confirm that new drives are compatible with current controller.
As you pleased with the way your server works - let it run. Just increase the capacity of drives. And in several years replace the server with new one (plus SSD configuration) if needed.
Again, better if you change the entire server adding SSD configuration and changing to Windows 2008 R2 that has automatic alignment to 2048 offset during installation. But not right now. Your current server seemingly has still resources needed for your business. And if it does fulfill your needs why to replace it?
0
 

Author Comment

by:jdjintx
ID: 33617621
"And if it does fulfill your needs why to replace it?"

That is pretty much what I was thinking, but I am here on the forum to make sure I do not miss any easy opportunities for system improvements when I replace the drives.  Maybe I am underestimating the complexity of replacing the SCSI drives with SSDs, but so far nobody has described any of the risks...  Before I close out this topic, can someone explain what problems could occur with option #3 above?
0
 
LVL 46

Expert Comment

by:noxcho
ID: 33617692
What problems? I wouldn't call them problems, rather complexities. You will need to take backup of existing RAID configuration then get a new RAID controller that supports SSD drive and configure new RAID.
Then restore from backup to new RAID configuration and adjust the OS to new hardware cause it will not boot from old configuration if this step is omitted.
Then you need to align the partitions of newly restore OS partition (and others) to Vista offset (2048).
And you don't know which of these steps could be more complex for you to accomplish.
0
 

Author Comment

by:jdjintx
ID: 33618131
"Then restore from backup to new RAID configuration and adjust the OS to new hardware cause it will not boot from old configuration if this step is omitted."

My OS is on a SATA drive that is not connected to the controller or part of the mirrored configuration for the SCSI drives I want to replace.  I expected that I could remove the controller & SCSI drives entirely (without replacing them) and have the server be unaffected other than not having access to the data that was on these drives.  I know this would work for any non-OS drive in any non-server type PC, but is this incorrect for a machine running Server 2003?
0
 
LVL 46

Expert Comment

by:noxcho
ID: 33619533
Hmm, your OS is installed on standalone drive and you want to replace the HDD with SSD drive?
0
 

Author Comment

by:jdjintx
ID: 33619588
My OS is on an internal SATA drive.  

Frequently accessed files are on mirrored 30G SCSI drives that are almost full.  

At a minimum, I want to replace the pair of 30G SCSI drives with higher capacity drives.

I am open to recommendations regarding drive configuration, as described in 1-4 above.
0
 
LVL 46

Expert Comment

by:noxcho
ID: 33619741
If data on pair drives is frequently overwritten then SSD is not good choice. Go on with SCSI or SATA drives.
0
 
LVL 11

Expert Comment

by:gmbaxter
ID: 33620536
In your situation considering that the server has served you well I would and seems to be doing a good job;

Find a 4 port SATA raid card that will fit your motherboard.

Buy two small SATA drives 160 gb or so
Buy two large SATA drives 500gb, 1 or 2tb

Create 2 raid arrays:
Mirror both 160 gb drives in raid 1 and put the OS on this.
Mirror both large capacity drives in raid 1

Partition the large capacity raid set so that you
 have 75 % for data storage and 25% for archive

this gives plenty of room for growth and adds OS redundancy.
0
 
LVL 95

Expert Comment

by:Lee W, MVP
ID: 33621197
Here's the thing - even if you were using a Dell or HP server class system, I'd be telling you to REPLACE IT.  ESPECIALLY replace a custom built system.

It's working fine now... Great... And it will.  Until it doesn't.  I can drive for years and not get into an accident... does that mean I shouldn't get insurance?  I shouldn't change the timing belt?  No... To ensure there are no problems, you have to take proactive maintenance.  Its your business... (job)... and you have every right to take chances with it... But as an IT professional, I think this is unwise.  Servers need to be protected so that when something does happen, you can correct it without causing yourself serious problems.

And you know what... MAYBE you'll keep using that server for another 10 years and nothing will happen.  And MAYBE I'll drive another 10 years and nothing will happen.  Things will fail, it's a question of when, not if.  And the more something ages, the greater the ODDS of failure soon.

I'll stop badgering you about doing (what I perceive to be) the right thing now.  I've offered my opinion on drive configuration and made my recommendations on your overall configuration.  Good luck.
0
 

Author Comment

by:jdjintx
ID: 33623131
"I shouldn't change the timing belt?  No... To ensure there are no problems, you have to take proactive maintenance."

That's where I'm getting stuck- you wouldn't buy a new car instead of changing the timing belt, and buying a new server rather than installing a larger capacity drive seems to be the IT equivalent of doing just that.  Am I that far off?


0
 
LVL 47

Expert Comment

by:dlethe
ID: 33623198
Yes, you are that far off.  Perhaps a better analogy would be buying a different engine for the car so it can go twice as fast.  
0
 
LVL 95

Expert Comment

by:Lee W, MVP
ID: 33624006
Thanks dlethe...

To clarify, You could drive a car 3x farther than the recommended changing time of the timing belt and be just fine.  Just like you can run an aged server beyond it's warranty and it can be just fine... Until that timing belt fails and now, instead of having scheduled maintenance, you are stuck on the side of the road with 3 screaming kids, a dead cell phone, a pissed off wife, and a pending car repair bill that is going to cost you FAR MORE than what changing the timing belt when you should have would have cost you.  And how much are those screaming kids and wife worth?  Always difficult putting a monetary value on things like that, but I'm sure you'd pay $10 so it wouldn't happen... $100?  $1000? At what point do you say I'd rather deal with the screaming kids than the do preventive maintenance?

The moral is that you can keep doing what you're doing, frankenstein a box, and keep on going with what you have and it may be just fine... but when it fails, will the pain caused by it's failure be minimal enough that it was worth your time going through hell reworking things?  I've had clients tell me "oh, that's not that important" and then "that's" not available and the world is ending.  Odds are better than a coin flip that you're going to be fine doing what you're thinking of doing... but the odds get MUCH better than you're going to fine with a new server quality system under warranty.  

Hey, if you're feeling lucky, hit a casino... don't take the chance with the network...

Sorry... Really, this time, I'm done.
0
 

Author Comment

by:jdjintx
ID: 33630450
 
Leew- no need for apologies; I’m here for the advice that comes with the answers to my questions, and frequently I’m not even asking all the right questions at the beginning of one of these sessions.  I am 100% confident that your philosophy is the best overall approach to server management.  Following your advice, however, costs money & even more time:  
  1. $2000-$3000 server + OS  
  2. Many hours configuring new server, unless there are migration tools I don’t know about (FAST?).  To the best of my knowledge, this will include:  
  • Re-installing our server-based accounting application  
  • Re-installing/configuring Kaspersky Administrator for internet security
  • Configuring DNS server (probably basic for IT guys, but for some reason I remember having a hard time with this last time)  
  • inputting user account details  
  • Reconfiguring the backup schedule  
  • Moving data files  

I agree that this is minor compared to losing Frankenserver altogether; Leew’s comparison to screaming kids, pissed off wife, and a huge bill is right on the money if it goes offline, but this is where I feel that my situation is not a good match for that particular analogy.  I can move all of our data (<30G) to a networked XP machine and avoid any downtime whatsoever.  Performance won't be optimal, but we ran like this for nearly two weeks after the flood without missing a beat.  Because I can avoid downtime, the cost & effort required to replace Frankenserver when I eventually need to are no greater than if I proactively do this right now.  If I buy a new server proactively, I prefer to wait until Q4 or Q1 when I have more free time than I have today.  Again, Leew’s advice is valuable & much appreciated, but I don’t want to invest the time right now if there is a practical alternative.  Unfortunately, I’m still a little bit confused by some of the comments regarding the alternatives and I’ll address that in a separate post…  
0
 

Author Comment

by:jdjintx
ID: 33630531
 
Barring a total server replacement, I still need to weigh my options for increasing drive capacity on Frankenserver.  From the comments that address drive changes, I’ve so far gleaned the following:  
  • Dlthe:  Approves of SSD’s in a server, suggests that in my situation I could even accomplish mirrored drives without a controller.  
  • Noxcho:  The simplest approach is to replace current SCSI’s with larger SCSI’s, but has concerns about properly restoring the OS.  SSD’s are not a good choice for drives that are frequently overwritten.  
  • Gmbaxter:  Replace the drives with four SATA drives on new raid controller, optimize OS/Data placement  
 
I think I’m pretty close, but I need a few clarifications to better understand my options & associated risks.  I’ve probably contributed to the confusion with my #$%^& terminology, so I’ll offer a clarification of my own.  I have four HDD’s in this machine, not four virtual drives:  
  • HDD1, SATA, connected straight to MOBO, has only the OS  
  • HDD2 & HDD3, SCSI, connected to PCI controller, only have documents/data, are mirrored (so I guess they count as one drive),  
  • HDD4, SATA, connected straight to MOBO, only has old data available to users on a read-only basis.    
At a bare minimum I want to increase the size of HDD2/3.  There have been several references to restoring the OS, but no explanation of why this would be necessary if I leave the HDD1 as-is.  
   
Dlthe:  I loved the car/motor analogy, but I’d like to take it just a little bit further…  I know better than to drop a HEMI in Mom’s Honda Accord.  However…  If Mom’s Accord is getting a new engine, we know that an identical 4-cyl should be plug & play.  Assuming the 6-cyl engine is also a fit for the frame, transmission, etc, it would also be plug & play, and we’d get better performance by opting for the upgrade.  Yes, a new car would be better…  Maybe they want to wait until a hybrid is available?    

Noxcho:  I know there are questions about SSD write endurance, but this article convinced me that "That myth - that flash SSDs wear out - now belongs to the past ", especially for as little data as I have at work.  Maybe these guys are quacks, but it seemed reasonable enough to me…  

GMBaxter:  It’s not clear whether your recommendation to replace all four drives is based on necessity to also change the drive with the OS or if this is simply a possible improvement.  
   
I think dlthe pretty much answered my question in his first post, it just wasn’t clear to me if new SSD’s could connect straight to the MOBO & be mirrored without compromising the configuration of HDD1 and HDD4 which were already there.  My follow-up post never got a direct response or we’d probably have been done a long time ago (DLTHE mentioned mirroring the SSD drives without a controller, which sounds great.  Can this be done in addition to standalone drives for the OS & archived files, or do all the drives need to be part of the raid/stripe/mirror configuration?).  I’m sure this was in part because of the confusing way I described the HDD configuration, but if dlthe could set me straight on that one point I’ll close thisout.  Otherwise..  

I know I could get answers to many of these questions with quick experiments, but I’m here on the forum to avoid the risks that go with finding out the hard way!  I think I’ll be OK if I can understand the following:  

  • With the OS installed on SATA HDD1, what would happen if I simply pulled the PCI controller & removed HDD2/3?  Would the server likely boot & function with HDD1 & HDD4 or would that likely compromise the whole system?  
  • If HDD2/3 can be replaced, I would prefer to leave HDD1 & HDD4 as they are and mirror two new drives connected directly to the MOBO without a controller.  Is this possible, or would that likely conflict with HDD1 and HDD4?  
  • It would be silly to upgrade HDD2/3 to SSD’s if improved speeds are not noticeable by the networked users. With Gigabit networking in place, should SSD performance even be noticeable?  
  • When a user accesses files on HDD2/3, does the speed of HDD1 (with the OS) have much impact on file access speeds, or is performance mostly dependent on HDD2/3?  
0
 
LVL 46

Assisted Solution

by:noxcho
noxcho earned 250 total points
ID: 33630629
With the OS installed on SATA HDD1, what would happen if I simply pulled the PCI controller & removed HDD2/3?  Would the server likely boot & function with HDD1 & HDD4 or would that likely compromise the whole system?  
-It must function if the OS is clearly installed on single SATA drive but not shared on two drives such as SATA and SCSI RAID. For example system files on SATA and boot files on SCSI.

If HDD2/3 can be replaced, I would prefer to leave HDD1 & HDD4 as they are and mirror two new drives connected directly to the MOBO without a controller.  Is this possible, or would that likely conflict with HDD1 and HDD4?  
-If OS drive is not touched then they will remain intact. Simply replace the controller or drives on your preferred choice. I still heard that SSD wears out specifically because it is working based on pages not sectors like your SCSI drives. Alignment can slow down this negative affection but still it will take place. And SSD drives are fast for READ operations not write. So this is worth of replacing for your HDD4 where READ only data is stored. Write operations will zero the advantage of read operations on SATA2/3. This is the main disadvantage of SSD drive for you here in my mind.

t would be silly to upgrade HDD2/3 to SSD’s if improved speeds are not noticeable by the networked users. With Gigabit networking in place, should SSD performance even be noticeable?
- I doubt. Mainly because of write operations are not as fast on SSD as read and secondly - because your users are going to access the drive via network and its load will affect the speed. So why spend money on SSD drives?

or is performance mostly dependent on HDD2/3?  
- Yes.
0
 
LVL 11

Expert Comment

by:gmbaxter
ID: 33630879
My recommendation was based purely on improvement.

You could stick with your single OS drive, but if that fails then there is no server, until you have rebuilt it and restored from a backup.

You could image your current server OS drive and restore it to the new OS drives on the sata raid card without having to re-configure everything.
0
 

Author Closing Comment

by:jdjintx
ID: 33631732
Thanks, I know it took a while, but now I'm comfortable with how this should work & what risks I have moving forward.
0

Featured Post

Give your grad a cloud of their own!

With up to 8TB of storage, give your favorite graduate their own personal cloud to centralize all their photos, videos and music in one safe place. They can save, sync and share all their stuff, and automatic photo backup helps free up space on their smartphone and tablet.

Join & Write a Comment

Hyper-convergence systems have taken the IT world by storm and have quickly started to change our point of view of how the data center should and could be architected. In this article, I’ll explain the benefits of employing a hyper-converged system …
Every server (virtual or physical) needs a console: and the console can be provided through hardware directly connected, software for remote connections, local connections, through a KVM, etc. This document explains the different types of consol…
This video teaches viewers how to encrypt an external drive that requires a password to read and edit the drive. All tasks are done in Disk Utility. Plug in the external drive you wish to encrypt: Make sure all previous data on the drive has been …
This Micro Tutorial will teach you how to reformat your flash drive. Sometimes your flash drive may have issues carrying files so this will completely restore it to manufacturing settings. Make sure to backup all files before reformatting. This w…

708 members asked questions and received personalized solutions in the past 7 days.

Join the community of 500,000 technology professionals and ask your questions.

Join & Ask a Question

Need Help in Real-Time?

Connect with top rated Experts

19 Experts available now in Live!

Get 1:1 Help Now