Server Upgrade on a Tight Budget

(EDIT: After reading your post about the server I'm rewording this)

I disagree about trying ANYTHING with SBS2003. It's just as outdated as XP at this point; if they're going to upgrade soon, it has to be all or nothing, and going "all" and falling back on an 8-year-old server OS is asking for trouble in the long-run. Furthermore, there's no need to run away if this place is treating and paying you fairly. They're clearly struggling with their infrastructure, so you should try to help them as best you can. But run if shit starts going downhill! This doesn't seem like a very complex network, but as your first job, it will be tough (I'm still in school for network engineering and on my first real job right now as well!)

IMO this is what I would do:

1. Formulate a long-term (say, one-year) plan with them. Go over budgeting. You will NEED to upgrade the clients at sometime in the future in order to do anything, so you need to go over that. Dual-core machines with 3-4GB of RAM and Windows 7 should do the trick. This is going to be the biggest cost probably, so it should be the last thing done.
2. RIGHT NOW, first thing's first, the server. I completely agree with virtualization. Get a SBS2008 license, and put that on the Precision server. Turn on HyperV and move the SBS2000 install to a virtual machine. Decommission the old server, it's junk. That will keep the server going for now.
3. If they are using hubs at all, they need a switch. You can get decent 48-port 10/100 managed switch for around $300-400 on Ebay, and decent non-managed Gigabit switches for a little more/same. They don't really need a managed switch, but they DO need a centralized switch. If you're going with a big company like Dell for (1), you can probably get this from them for cheap too, but this is much more important than the client computer upgrades.
4. Backups can be moved into the Precision. The 1TB external (put it in the machine or in a hot-swap enclosure for easy removal; this is the "disaster: grab it and run" disk) + another two 2TB drives for internal backups/data.

That's all I can think of right now, and that's how I'd personally handle it, but since I don't know the client I can't say for certain. Good luck!
 
Last edited:
The proplem is I do not think you can upgrade from SBS 2000 to SBS 2008. And even then SBS 2011 is suppose to be coming out.

I would not run the Precision as a hardware server as you will then have to migrate it to something else. I think you are going to be better off with ESXi and making the SBS 2000 a VM. Atleast then you can move that to better hardware in the future.

You will want to see if you can do an upgrade from SBS 2000 to SBS 2008 or 2011. If not then you will be doing a new domain, which you can start in VM as well.

If you can't do a straight upgrade from 2000 to 2008 or 2011. What I would do is get a cheap copy of SBS 2003, upgrade SBS 2000 to 2003 in VM. When they get funds, purchase a new server Dell T410 or T610, and then do a SBS 2008/2011 upgrade to the physical hardware, and at that time pick up some Dell Vostro or Optiplex.
 
marley1 said:
The proplem is I do not think you can upgrade from SBS 2000 to SBS 2008. And even then SBS 2011 is suppose to be coming out.

I would not run the Precision as a hardware server as you will then have to migrate it to something else. I think you are going to be better off with ESXi and making the SBS 2000 a VM. Atleast then you can move that to better hardware in the future.

You will want to see if you can do an upgrade from SBS 2000 to SBS 2008 or 2011. If not then you will be doing a new domain, which you can start in VM as well.

If you can't do a straight upgrade from 2000 to 2008 or 2011. What I would do is get a cheap copy of SBS 2003, upgrade SBS 2000 to 2003 in VM. When they get funds, purchase a new server Dell T410 or T610, and then do a SBS 2008/2011 upgrade to the physical hardware, and at that time pick up some Dell Vostro or Optiplex.
Click to expand...

You would need to migrate the domain through 03 before going to 2008. At least with sbs. With how this place sounds like it is setup I'd prob just build a new domain and be done with it.
 
buying a 48port switch hub isn't going to help his wiring needs when they are using hub's to extend lines, it would think he needs a re-wire in certain locations of the building.
 
The cabling problem is a simple fix. All workstations are within a 250' cable run of the server room, so new cable just needs to be laid, and the central switch needs to be replaced with a switch with a higher backplane. The cabling is the lesser of the evils, so fixing it is a ways off.
 
Alright, I hypothetically choose to upgrade the memory and HDDs in the Precision 690 to act as a temporary ESXi server. I'm not too familiar with memory configurations on multi-socket mobos. Based on the fact that the all four DIMM slots for CPU0 are full and the slots for CPU1 are empty (current mem config in the Precision) combined with what I know about desktop mobo memory configurations leads me to believe the following:

1) The DIMMs located next to each CPU on the physical board create a logical pool which cpu's draw from, hence one set of DIMMs can be full and the other empty.

2) Like desktop memory,the (using a dual socket example w/ 4 DIMMs per cpu) DIMMs on CPU0 have a corresponding slot on CPU1 that must have the same DIMM size. So CPU0 DIMM1 would have to have a memory stick of capacity equal to the stick in CPU1 DIMM5 (cpu0 has DIMMs 1-4, cpu1 DIMMs 5-8), and so on. DIMM2 -> 6, etc.

I think this is right. The machine currently has 2 gigs (512mb x 4) and as long as I mirror the DIMM configuration on each CPU, I can install say another 4 gigs to keep costs down. So DIMM: 1 - 1gb, 2 - 1gb, 3 - 512mb, 4 - 512mb for CPU0 and DIMM 5 - 1gb, 6 - 1gb, 7 - 512mb, 8 - 512mb.

I just want to double check since I may be off, but
 
Striker109 said:
Does your client have a budget in mind? Saying "tight budget" could mean $500.00 or $5000.00.
Click to expand...

At the moment, it's around the $500 range, if that. He knows that his network is unstable right now, and I've reiterated many times that any fix short of relatively new, proper server hardware must be viewed as a temporary, last resort solution. At least he seems to understand this, and I will go over it again when I present the final options to him.

I will, of course, keep the old server around as a fallback in case of catastrophic failure.

I was able to find a version of ESXi that is fully compatible with the Precision, and if it looks good after testing, I'll add more more memory and a couple HDDs to it. At this point, a P2V conversion seems much safer, quicker, and easier than migrating the OS to a slightly better hardware solution.
 
$500 for the project, or $500 for the server upgrades? I wouldn't even run the cables for that...
 
aaronearles said:
$500 for the project, or $500 for the server upgrades? I wouldn't even run the cables for that...
Click to expand...

$500 for this specific "upgrade." He's trying to spend as little as possible on hardware. My fees are separate from this budget, as he needs someone to take care of the network regardless.

I got off on a tangent with the cabling and other issues. Those issues will be visited later on when business picks up.
 
ebay the ram you can get some good deals on therecould probably buy 8 1 gig sticks for like 100-150$ buy a second cpu kit & heatsink ( they are kitted )

Change the drives too, don't use older hdd's upgrade to new 1tb satas or 2 tb satas at most they are 89$ each.

Or use the kingston memory finder and see what they charge per stick, im going to assume its ddr400.

Take the asset tag to the dell support center, it will tell you exactly what the specs of the machine is, you could also type in the machine model on ebay and you an see cpu & heat sink combo's and ram packages, make sure you get the same cpu as what you have in there now tho :)

j'
 
OH and if you don't have a tech-net account GET ONE! lol :) best 200$ you or your boss spent.
 
dashpuppy said:
OH and if you don't have a tech-net account GET ONE! lol :) best 200$ you or your boss spent.
Click to expand...



The first thing I did was search the internet for a TechNet discount code so I wouldn't have to pay full price. That is the lifeblood of any techie. If anything, the TechNet account will come in handy as I'm currently in the process of making a large VM network to refresh myself on server 2003/2008 and misc server software.
 
Last edited:
aaronearles said:
I hope you're not recommending that he purchase a technet subscription and resell the licenses to his client...
Click to expand...

why would i do that ? it states in BIG bold letters SOFTWARE NOT for resale when you check that you have read the ULA when signing up with ms technet.
 
Well he's talking about upgrading a client's OS, and you threw out the Technet suggestion, I'm just making sure he knows not to do anything stupid.
 
i know with actionpack you cant really use it for business, dunno if technet is same.
 
DeChache said:
Pretty sure Technet is only for training and testing no productions environments.
Click to expand...

You are correct.

http://technet.microsoft.com/en-us/subscriptions/default.aspx said:
Software is licensed for evaluation purposes only-not for use in production environments. TechNet Subscriptions include the most recent Microsoft software version. Visit Microsoft Software License Terms for details on your use rights for evaluation software and other components of the TechNet Subscription product.
Click to expand...
 
I hate to add fuel to the off-topic fire, but how much to technet and MAPS differ?
 
I have a couple of more issues for which I could use some experienced opinions to supplement the my own research.

Server
I've considered all the input and weighed all my options, and have made a decision. I feel in this situation, adding 4GB to 8GB of RAM (4 plus the current 2 for 6GB or 8) a SAS HDD to the Precision 690 and going the route of OS virtualization is the right move. Of options considered, I think it is the lesser of the evils for a few reasons:

1) A P2V conversion of the server will be much faster, easier, and safer than performing a hardware migration.

2) Between looking for a cheap server ($500 max) and adding some RAM and a SAS HDD to the Precision, upgrading the Precision is the better option. Spending roughly $400 or less - depending on where I buy the memory and HDD(s) - now to make the Precision a reasonable last-resort placeholder has a better chance of giving the OS a more up-to-date hardware platform than buying a used server for the same price. I know this removes some options for HDD fault-tolerance, as a used server may come with a RAID5 array filled with SCSI drives, but that can be mitigated through revamping the backup plan and keeping the current server around a a backup (after a thorough cleaning, both physically and digitally).

3) The Precision uses server-grade hardware (Xeon CPUs, fully buffered ECC ram, and a mobo designed for high-performance applications). A problem with it revolved around finding suitable hardware drivers for a server OS, as dell only wrote drivers for client OSs. Since I found a free, fully compatible version of ESXi, that issue no longer exists.

4) Planning the network environment around virtualization will benefit this business greatly in the future to help the network become more cost-effective.

After spending a day testing ESXi, I have to say I'm beginning to become a fan of virtualized computing. Now that I have this problem resolved, I have some questions regarding common practices for designing and implementing a backup plan to a network.

Backup Plan
I have only had time to make minor adjustments to the current backup plan, such as creating a new routine for the exchange server that actually has a save path. Currently, there are three scheduled backups using ntbackup. One routine backs up the dialer control server, one backs up the exchange database, and the other backs up the DC in its entirety. All backup routines are normal backups that run 5 days/week (M-F) and send their completed backups to a 1TB external HDD (after I made sure no backups were being saved to the computers being backed up). I'm trying to design a new pan that is practical for a small business. I have some ideas as to what I can do, but could use some input for certain aspects of the plan. The financial situation is the same as the server, so I'm trying to make an improvement using what is already available, including the offsite backup service included in my clients internet and phone package.

I have the following plan in mind for a one week backup schedule:

1) Full backup on Sunday. This would be a normal backup type that would overwrite the old backup file every week.

2) Incremental backups M-F. Each backup file would be overwritten each week by its corresponding day.

3) Upload a backup to the ISP data center once per week. I feel like a full backup should be uploaded during off hours the day after it is created, but considering there are two T1s available for internet and some of the current backups are very large (30GB or more), this route would not be cost-effective. See questions below for more details.


Questions regarding common backup practices:
1) For a small business that deals in sales, would a one week backup schedule be sufficient or should it cover 1 month back?

2) The two data stores that are essential are the exchange first storage group and the SQL database. All payroll data is already stored offsite. I would think that the most critical data to upload is the SQL database as it stores all the customer information for the contact management program. The exchange storage group can be backed up on the external until a more permanent solution is financially viable. Thoughts?

3) In addition to a full backup, is it worth creating a separate backup routine for the exchange storage group? My gut says no as that would be redundant as the backups are on the same storage device.

4) Are there any commonly accepted methods of incorporating snapshots of the virtual machine image into the backup plan?

5) Any other recommendations here or via PM would be great!

Network Documentation
This also leads me to another important issue: documentation. Since there is none, some needs to be written, and I can do so while I deploy the band-aid server. I found some sites such as networkdocumentation.com/, some tools for gathering OS, Exchange, and SQL server info like SYDI scripts, and some free programs from solarwinds and other companies. I want to gather this info and keep the more static configurations in a binder for reference to make troubleshooting easier.

I'm interested in learning what tools are commonly used for gathering network information and how much data network data is collected for documentation, as well as how that data is formatted and kept once collected. I'm doing my research and have collected some tools, but I'm sure there are some great resources I haven't found on my own yet.

On a specific note, if there is anyone who is familiar with predictive dialing systems, please PM me, I have a question about configuration.
 
Last edited:
Berg0 said:
I hate to add fuel to the off-topic fire, but how much to technet and MAPS differ?
Click to expand...

What is MAPS? I've got access to both MSDN and Technet, and as far as software availability and interface, they are the same. Not sure if there are subtle differences elsewhere, since I go straight to the downloads page!
 
To my understanding, technet is more geared towards the individual, while MAPS (Microsoft Action Pack Subscription) is geared more towards a Microsoft Partner to give them software to run their business. From what I've seen MAPS has much left software than technet but it has the essentials for what you should need.

As of a few months ago, MAPS now comes with 1 technet subscription.
 
Your wasting your time. If they are not willing to spend over $500 to replace a 10 YEAR OLD SERVER then they do not take their network seriously, and neither should you. Your plan has some flaws, major ones.

First, the operating system hasn't been officially supported by Microsoft for a few years. No security updates, patches, ect.
Second, your hypervisor has no redundant disks. You realize that if you lose that disk your done, right? Data backups are great, when your restoring backups to modern operating systems. Whats your disaster recovery plan? Why not drop the sas drive and get 2 sata's and a hardware raid card? Your data store is your primary concern, not something you want to skimp on.

Third, are you sure a P2V converter will work on such an old OS? What if it doesn't, then what?

Fourth, lets say you do pull it off, and then that used percision dies and they lose data or are down for an extended period of time. Who is going to take all the shit? You are. That brings me back to my original point. If they are not willing to spend the money on the right equipment, then run away.

Yes, I am the untangleappliances.com dude, but I ran (and still do) a consulting firm in Pittsburgh before I started untangle appliances. You have to know how to handle cheap bastards. I tell my clients what they need. It is not open to negotiation. If you want me to take responsibility for your network, then you buy what I say you need, nothing less. My suggestion to you would be this. Get a real server and at least raid 1. If they are not willing to spend a grand, then bail. I know it sounds crazy, but these kinds of customers are not worth the risk and hassle.
 
^ This.

That's really back to reality about the whole situation.

They're not willing to take their IT needs seriously and while you want the job and the experience, it's all going to fall back down on you hard.

Either make him see your way about what really is needed or you have to bail.
 
Comicbob, This is the kind of job that can come back to haunt you and your career. If you choose to go forward you must be very clear ON PAPER and verbally to cover your butt. That way if stuff hits the fan you can just bring up the paper that he signed. If he refuses to sign this document walk away.

That said, I wouldn't run the old hardware or put anymore money into it. Since he's just throwing money away.

There's some things to consider as well...
- monthly payments as suggested earlier in the thread will help the boss's wallet feel less strained. Maybe sell the older hardware to recoup some of the cost of the new server.
- If this is the primary server. Redundancy is a MUST. You won't have the amount of redundancy that you'd need with the precision. A true server in a working environment needs Redundant power supplies and raid 5 or 6 (depending on the number of drives). Better yet keep one of the servers/workstations around as a fall back in case the worst case (motherboard failure/mother nature) happens to the primary server.
- Get a server with a good warranty and next day service. something along the lines of a 3-5 year plan will help both the boss and you in case something happens. Also piece of mind. :)

Just tell him you researched it alot deeper and wouldn't recommended going forward with reusing the old hardware for anything more then maybe backup servers.
Getting it done right now will cost him less in the long run (6 months to a year down the road).
 
Proactivens said:
Your wasting your time. If they are not willing to spend over $500 to replace a 10 YEAR OLD SERVER then they do not take their network seriously, and neither should you. Your plan has some flaws, major ones.

First, the operating system hasn't been officially supported by Microsoft for a few years. No security updates, patches, ect.
Second, your hypervisor has no redundant disks. You realize that if you lose that disk your done, right? Data backups are great, when your restoring backups to modern operating systems. Whats your disaster recovery plan? Why not drop the sas drive and get 2 sata's and a hardware raid card? Your data store is your primary concern, not something you want to skimp on.

Third, are you sure a P2V converter will work on such an old OS? What if it doesn't, then what?

Fourth, lets say you do pull it off, and then that used percision dies and they lose data or are down for an extended period of time. Who is going to take all the shit? You are. That brings me back to my original point. If they are not willing to spend the money on the right equipment, then run away.

Yes, I am the untangleappliances.com dude, but I ran (and still do) a consulting firm in Pittsburgh before I started untangle appliances. You have to know how to handle cheap bastards. I tell my clients what they need. It is not open to negotiation. If you want me to take responsibility for your network, then you buy what I say you need, nothing less. My suggestion to you would be this. Get a real server and at least raid 1. If they are not willing to spend a grand, then bail. I know it sounds crazy, but these kinds of customers are not worth the risk and hassle.
Click to expand...

I'm not sure why I didn't choose disk redundancy. Stupid mistake on my part. I'll learn from it though.

I've checked OS compatibility for ESXi and VMware's converter and they both list 2000 SP4 support. More testing is needed to be sure.

If anything goes wrong at all during the migration, I plan to simply stop and restore everything back to its original state.


pwrusr said:
Comicbob, This is the kind of job that can come back to haunt you and your career. If you choose to go forward you must be very clear ON PAPER and verbally to cover your butt. That way if stuff hits the fan you can just bring up the paper that he signed. If he refuses to sign this document walk away.

That said, I wouldn't run the old hardware or put anymore money into it. Since he's just throwing money away.

There's some things to consider as well...
- monthly payments as suggested earlier in the thread will help the boss's wallet feel less strained. Maybe sell the older hardware to recoup some of the cost of the new server.
- If this is the primary server. Redundancy is a MUST. You won't have the amount of redundancy that you'd need with the precision. A true server in a working environment needs Redundant power supplies and raid 5 or 6 (depending on the number of drives). Better yet keep one of the servers/workstations around as a fall back in case the worst case (motherboard failure/mother nature) happens to the primary server.
- Get a server with a good warranty and next day service. something along the lines of a 3-5 year plan will help both the boss and you in case something happens. Also piece of mind. :)

Just tell him you researched it alot deeper and wouldn't recommended going forward with reusing the old hardware for anything more then maybe backup servers.
Getting it done right now will cost him less in the long run (6 months to a year down the road).
Click to expand...

As much as I want to do this, and I think I can, the comments about making sure I cover myself have me thinking very hard if I even want to pursue this anymore. It feels like there is a clear understanding that this is may not work, but you and everyone else saying walk away may be right, even if I don't want to. I planned on writing up a plan covering what I'm going to do and how I'm going to do it anyway, but it may be a good to include a section detailing the risks and get his signature.

I've suggested financing to him for an upgrade, that was vetoed, so I guess that should have been a clear indicator that I should not get into this.

The advice of just dropping this job seems to be outweighing the positive feedback. I have a learned a lot and have a lot to consider right now over the next couple days. My client knows full well that his network needs work, but when it comes down to it, his inability to do anything about it financially may be the same thing as a lack of willingness to figure something out.
 
Last edited:
The advice of just dropping this job seems to be outweighing the positive feedback. I have a learned a lot and have a lot to consider right now over the next couple days. My client knows full well that his network needs work, but when it comes down to it, his inability to do anything about it financially may be the same thing as a lack of willingness to figure something out.
Click to expand...

If he KNOWS his equipment is on its last leg, and it IS, and he won't allocate any more than $500 to it, then he doesn't think it's important. Walk away, and let HIM deal with the fallout.

Do a detailed workup for what NEEDS to be done (1. new server, 2. new networking equipment) and the costs. Give it to him, and tell him it's the only way. IF he refuses to see the light at this point, tell him you're sorry but that you're not here to help throw money at an outdated and frankly patchwork garbage network setup. If he's not willing to take his IT infrastructure seriously, there's not much you can do to help him, because that's the first step: he needs to see that this is IMPORTANT.

EDIT: scotty_do hit a few nails I misses, esp. having them call you. I'd be frank and ask him: "How much does one week of server downtime mean to you dollar-wise?" That should be (at least) his budget to get this mess fixed. Because when it fails, I assure you it'll be more than 1 week of downtime...
 
Last edited:
He's bang on. I didn't want to say anything to get you discouraged on your first project. IMO running ESXi on that box will only make things worse. The overhead is too much for it to handle.

Take some time, put together a detailed quote and proposal, even explain to them what everything is. When they are willing to budget for everything, have them call you.

In the mean time, get them to spend $500 on verifying backups, or implementing a proper backup solution....they will need it soon enough.
 
scotty do said:
He's bang on. I didn't want to say anything to get you discouraged on your first project. IMO running ESXi on that box will only make things worse. The overhead is too much for it to handle.

Take some time, put together a detailed quote and proposal, even explain to them what everything is. When they are willing to budget for everything, have them call you.

In the mean time, get them to spend $500 on verifying backups, or implementing a proper backup solution....they will need it soon enough.
Click to expand...


The entire purpose of this project is simply getting the server to a state where it doesn't constantly crash every morning so it can undergo a proper upgrade in about 6 months. It's also about making sure it's done correctly, so if it turns out that it can't be completed guaranteed to work, then I won't do it.

I've thought about this heavily over the last day, and I've decided to stop and weigh the risks to me and to his business versus the chance of a successful migration. On paper my plan looks doable, but without a financial safety net, I guess there is no guarantee of success, which means a complete overhaul is the only option to bring his network any real measure of stability.

This has been an extremely valuable experience for me as it's a great example of how a company can screw itself if they don't keep their hardware and software updated. At least I have now a crystal clear example of what hitting rock bottom with a network means.

So, at the moment, I'm now heavily considering just leaving this alone.
 
Last edited:
What does this company do?

They dont have credit? You can get everything leased in payments with teh $1 buy out. Sure you will pay more but maybe the monthly payments will make it work.
 
marley1 said:
What does this company do?

They dont have credit? You can get everything leased in payments with teh $1 buy out. Sure you will pay more but maybe the monthly payments will make it work.
Click to expand...

It's a new, small telemarketing company. I mentioned financing already, and that idea was shot down immediately. I can mention leasing. It may look more feasible - I can't say for sure. I'm pricing out what it would cost to make the server and desktops current and am going to enter that into the quicklease payment estimator. That's assuming his business software is even compatible with current OSs.
 
ahh, gotta see.

dell vostro are good deals
dell t410 or t610

dell has a direct lease team.
 
Get them setup with a dell line of credit and get everything done in 1 shot. Sure, it is expensive, but the ROI will be in a much shorter time frame.
 
That thing is basically exactly what the Precision 690 that he current has is but without the rackmount case, the RAM is the cost prohibitive issue in fixing up either of these machines to the level they'd need to be to handle the overhead of this, unfortunately the boss already shot down that option once before, so..

He's writing up whole proposals, stuff that needs to be fixed and requires much more significant funding than he'd like but needs to be done, and is trying to make him see his way, but if he turns this down or even in phases or with different funding/leasing/credit options, it'll be time to get out of it..
 
If they are a true non-profit the smartest thing you can do is get them setup with techsoup.org and ccbnonprofits.com.

With these two you can get a majority of the software and hardware you need for pennies on the dollar.
 
You must log in or register to reply here.
Back
Top