After getting approval for a new test server after my 2003 server became inadequate towards our goals, I have been given a new server in which to use Server 2008! I don't have it all put together yet, but I am sure I will have some questions once I start using it. I am creating this thread as a starting point, and perhaps some of you will be interested in seeing the development of this project as well. In this thread you can comment on the ideas we have or the hardware we are going to be using, or ask the usual questions I get such as "why would you do such a thing"...

Anyways, here is the hardware we have lined up:

OS: Server 2008 64bit
MB: Intel SE7520BD23
CPU: 2x Xeon 2.80GHz dualies
RAM: 2x 2GB (4GB) Crucial C35DR260
Intel Pro/1000 MT DUAL PORT nic

RAID: 3Ware 9550SXU-4LP + 2x WD 80GB SATA for mirrored system volume.
RAID5: 3Ware 9550DX-8LP + 6x WD 400GB SATA for storage

We'll see how this server board turns out. Our big problem might be that Intel has no plans to qualify this hardware for 2008. Let me know what you think of this setup, or possible issues I may run into, besides the obvious motherboard drivers problem.

What couldn't you do in Server 2003 that you can only do in Server 2008?

Server Core, TS Gateway, TS App Publishing, Hyper-V Clustering, Read-Only domain controllers, ADFS/ADCFS, WDS in multicast, NAP for clients to start (all of which I use heavily).

what are the plans for the Server?
What Role is it going to play in the environment?
What applications and depedencies is it going to have?

lots of questions to ask

also i would worry about using this server if it is not going to be approved for 2008. That means you are going to have to fight drivers that aren't validated for 2008 and with the re-designs that happend you might be causing yourself more hassle then it will be worth

The WDS in multicast is the #1 reason why we're moving to 2008. I am sure that all those other things are going to be really awesome too!



The following roles will be assigned:
-Primary Domain Controller
-WDS
-Fileserver
-PXE Boot Server

This is just a test server to start. Our target for production (which is 2008 certified) is currently in use so I can't stop production just to play around on it. Because of this, it is also not a huge concern that I will have to fight with the drivers or settings to get it all working, as this will be a learning experience and allow the actual migration to go smoother. Also, because this machine is kinda bulky to be used only as a test server (mainly the 1.45TB of storage it has) I will be planning on changing the roles to match the following:

-backup domain controller
-dhcp server
-backup file server
-backup (ie technician PC) WDS

While the current production server will handle PXE Boot, WDS (Primary) and image and file storage.

Currently our test environment (which actually does have production value in certain instances) runs in the following configuration:

Current Config
1. Server 2003 Standard x32 (low end) Domain Controller with WDS, SQL Server, IIS, DNS, PXE
2. Server 2003 Standard x32 (high end but broken) running DHCP

The main problems faced with Server1 is that it is not spec'd for running the OS, and can only image 4 computers at once. The main problems with server2 is that it is supposed to have 1TB of space, but one of the drives fell off the array. It was retired (and destined for the recycling center but I saved it because it has drive images for legacy systems that our repair department can use. We also had tried getting those images off of the drive onto an external or network drive but the server would either BSOD, reboot or lock-up. Since it still kinda works, I use it for DHCP only and joined it to the domain.

This test server will replace server1 in testing, and after testing is completed, will replace server2 and enter the production environment.

Wow. You guys must be in H Core enterprise environments...nice...

Yessir.

Oh, I put in a request to get a rack unit to be put in the production area. It will hold 3 servers, the primary, the backup and the legacy with the busted raid. All in all, we will have about 3TB of network storage!

Update!

On Friday, I got fed up with the DHCP server and the fact that the mouse wasn't working... So I decided to reboot it. Well on the way back up, another drive fell off the array and the two remaining showed up as DEGRADED. I didn't want to deal with it then so I shut off the monitor and left for the weekend. Today I came back in and somehow the server recovered and I logged into Windows. It still didn't recover the 4th drive but it must have rebooted over the long weekend and found the 3rd drive that fell off on Friday. I am lucky for that because I was still able to use WDS and an RMA came in and that server was the only place that had the image I needed.

Today, the new server was set up with 2x 80GB mirror for the OS drive. Server 2008 Standard x64 was installed and it took about an hour. Then the 6 drive, 1.45TB RAID-5 was built. Those few steps took about 4 hours, and now the server is waiting for me to set it up.

I haven't came up with an idea for a domain or computer name yet.

I do have some questions regarding the migration I need to do. Here is the situation:

I have my 2003 Server (computer named UNCLESOCKS) that is a domain controller and the domain is called SHARK. Now I am pretty sure the new domain won't be SHARK but I might keep it if it is easier. What I need to do is take the new server and make it the top-level domain controller, and then take UNCLESOCKS and set it up to be the backup domain controller. So here are my questions...

1. Is there a tutorial on how to switch a server role from the primary to backup domain controller?
2. Is it important for me to unpromote UNCLESOCKS before I promote the new server?
3. Would it be easier to use the domain SHARK rather than starting fresh?
4. Is a 2003 Server capable of being a backup domain controller to a 2008 Server?
5. Does Windows 2003 have the ability to automatically backup files from another computer, using a differential or incremental synchronisation?
6. If yes for #5 (or with a 3rd party product) Would there be a problem if the 2003 Server backed up data from the 2008 server?

As far as the AD portion is concerned, I have not created any additional OUs and only 2 user accounts on the SHARK domain, so I would not need to worry about AD migration from one server to the next.

I will also like to find information regarding the virtual desktop/terminal emulation using virtual OSes within 2008 that our Microsoft rep told us about.

Server 2000+level domains don't have "primary" and "backup" domain controllers. DCs are DCs, there's no differentiating them (other than RODC). The only thing you need to know is where the various "role masters" are (Global Catalog, Infrastructure master, etc.).

If you plan to keep your current domain level, there's no "unpromoting" to do. You put the 2008 server in the domain as a DC, move any role master stuff off the 2003 box, then you can just shut it down. You *could* completely remove it from the domain by demoting it but it's never a bad thing to have a DC ready to go in an emergency. 8)

Again, if you are keeping the same functional domain level, there's nothing to change. Add the 2008, remove the 2003. The network should not be affected. If you intend to move to a native Server 2008-level domain, then you would need to make a brand new domain.

Should you decide to make a brand new Server 2008-level domain then you will have issues getting Server 2003 boxes to do stuff in that domain. Ideally a Server 2008 domain is ALL Server 2008(+) DCs.

There are some little things about 2008-level domains (and forests) that make them appealing, but for smaller setups those things aren't very important. There's some network optimizations for the databases and stuff but overall a 2003-level domain is fine for small and even medium-sized domains still.

The idea ATM is to not remove the 2003. I currently use 2 2003s for the setup, the idea is to run 2008 as the main AD DC and the 2003 as a backup and run DHCP only. Removing the 2003 is not a priority right now. The other (the one with the busted RAID) I wish to remove from all domain activities because its failing as it is. If that server weren't so *blasted* important it would have been in the dumpster over a year ago.

Eventually, the setup will be two 2008 x64 servers but for now I want to learn partial migration for them. Aside from my own learning, it may be useful information for our clients.

EDIT: sorry didn't know about the swear filter lol.

Staying with a pre-2008 level domain makes things easy. Just add the Server 2008 box as a DC. Like I said, there is no "primary" DC, no "main" really. If you want you can throw the master roles on it (I don't recommend global catalog and infrastructure on the same box) and be good to go.

I have to say that one of the hardest parts of this project is that upper management thinks that we (the guy who handles the REAL server and myself, who handles the product testing/development) can get the new server system up in running in no time. He has no idea that we will have to do tons of testing first... uggg...

Today I actually did some more work. On the 2003, I added the new server (GEARS) to the Domain Controllers group, and added a domain user (SHARK\GEARS) to the Domain Admin group. Then I joined Gears to the Shark domain, and added the roles of AD DS and WDS. I also added some functions such as Backup Services. To my disappointment, I was unable to add Hyper-V because it said the CPU was not supported, even tho the server has the 2 dual core Xeons in it... What exactly does Hyper-V require anyways? I also formatted the RAID-5 and was pleased to see the 1493GB listing there in Disk Administrator!

I still have a bunch of notebooks to image using the 2003 as the WDS, so my role migration will have to wait until those are done...

I do have a question regarding my config with the 2003 server (UNCLESOCKS)... here is the setup:

Domain Controllers: GEARS (Computer)
Delegated Control for active directory functions to GEARS (Computer)
Domain Admins: GEARS (User)
but when I log into GEARS, I use the Administrator account. Should I add GEARS\Administrators into the Domain Admins group instead? When I log into GEARS, it doesn't give me an option to log into a domain, or is there another option I should be enabling?

http://www.microsoft.com/windowsserver2008...icalInformation

You need Xeons running Intel's VT from this list (Select "Intel Virtualization Technology" from the "Supported Features" list and click the "Filter on Selections" button). I am not sure if it's a guarantee that everything on that list will work, but it's for sure anything not on that list isn't .

This would be an excellant time to look into Microsoft Deployment ToolKit. Using this and the MultiCast Feature in 2008 you could image all your laptops in a relatively short time


When logging onto a domain Controller you should only recieve the option to log onto the domain (since there are no local user groups on a DC) you can log onto the DC with any account in the Domain Admins group though.

Perhaps there is something I need to enable in the BIOS, or install from Intel's website so Windows sees that it is capable of running it. We are going to need to use Hyper-V to support our legacy customers (ie those who require 16bit Ghost to be used).



I am already using that, but we can only use the deployment features provided by the OPK toolkit, not the "corporate" toolkit. Right now, these notebooks can image in 3 minutes using the 2003 WDS. So with this relatively short image time, it is not a problem to do one at a time, since we don't get that many orders for that notebook model at once. But, using WDS on the 2003 for production is not the original idea, its just that some network controllers (software controlled like Broadcom57 and Realtek 8111B) do not work with our current imaging system.

In other news, the 2003 server started acting up ever since I joined GEARS to the domain. It now sees our legacy server as being a 'rogue' DHCP server, even tho it is available as a managed DHCP Server in UNCLESOCKS' DHCP management control panel.

Was it re-authorize the DHCP server to refresh the records on it

Well I am thinking that it is getting this error because DHCP on UNCLESOCKS is disabled, and I am authorizing the legacy to run DHCP instead. I did it this way because I couldn't get any computers to boot via PXE with DHCP and WDS running on the same computer.

Well here's some bad news...

Cluberti might better understand this situation.

See the original plan was for me to create a test server, then a second server for production was going to be built. In fact, we had already gotten in the parts for the new server, sans (no pun intended) the hard drives.

An executive decision was made and in order to save on the system costs (we are getting one next to nothing after we requested eval products from Intel) they would take the hard drives in my test server.

So basically the powers that be have scrubbed my project. They are taking the chassis, one of the RAID controllers, all 8 HDDs and the PSU, and MAYBE the CPUs from my project so I'm basically up the river without a paddle.

Thanks a lot for all of your help and input! I am sure I may dream up another project to make a thread about in the future!

Gotta love corporate priorities... I work in the IT department at my place, and we were asked a few months ago to prep a couple of very powerful workstations for another department. Afterwards these PCs got lost in red tape and ended up stuffed in a closet. We "found" them the other day and set them up in our area. I fully expect someone to walk by, see them, and re-allocate them. They are cool, but I am not getting too attached ; ;

yeah, you have to fight and show the business that what you are doing will have an huge impact down the road. It is all dollars and ROI to them, if you can show that testing now, will save you tons in troubleshooting, incedents, and lost productivity time. You should be able to convince them that you should be able to get the parts you need

Well the parts were taken to build a new more compatible 2008 server, except that this project is being transferred to someone else too. I think I'll still get to manage WDS on it once it is up and running but who knows.

For those of you who are running Windows Server 2008 now, I noticed in the 'help' files regarding enabling shadow copies, that you are now able to store them on a different volume. This was not possible in Server 2003. Has anyone tried this?

Might as well be running it on eeePC

So far I am still waiting to get my hands on the 8GB EEE PC. I've only gotten the 4GBs so far... They work with XP Pro and can only be imaged with ImageX and not the GHost we use. They are cool little things for sure.

I've got Vista SP1 , VS2008 Team Edition and Photoshop CS3 on my 4GB 701 and it works satisfactory. Server 2008 did just as well.

Yes, i have the similar problem.

Actually, I am a small business consultant, and one of our clients is a small architect firm. One of their servers is a Windows Server 2003 x64 box with 8 GB of RAM and 2 TB of storage. Their data usage is NUTS! Another server is SBS 2003 primarily for Exchange and a few other things. They have an FTP server and they also have a server dedicated to ISA 2006 and their plotters.

I've been doing a bit of work with our new 2k8 server, as the guy it was assigned to doesnt know how to use WDS.

Anyways, we just upgraded it to 16GB of RAM so it runs real nice. We also are running a 1.7TB RAID 10 for storage, and some smaller capacity mirror for the OS.

Right now we are battling this issue:
http://www.msfn.org/board/System-Error-123...ve-t120006.html

Also, right now we are experiencing a lag issue but can't seem to trace it. The server is operating with 2 teamed (Team 0) gigabit NICs using Cat 6 cable on a gigabit switch. When we are deploying an image, network utilization is average 95%. When capturing an image, util is 2% or less, with spikes up to 40%.

Doing a comparitive study, this means on a capture the 2008 is getting an average of 87Mbps from a client on a gigabit network. Compared to my 2003 server which gets an average of 58Mbps on a capture on a 100Mbps network. While 2008 network util is 0.40% to 2%, the 2003 gets 43% to 76%. But these numbers lie. Both servers are experiencing approximately the same capture speed, but experience different deploy speeds.

What could be causing this limit on the capture (aka data-in) speed?

I've seen the same problem too, I'm guessing that the RWMs used by the WDS are still using the previous (6.0.6000) version of the wimgapi internally. Capturing manually with latest ImageX doesn't cause this problem, or at least not to the extent you're describing.

AFAIK, when Mr. Bangalore Guruprakash (brGuru) first implemented the WIM algorithms they were not designed to handle UNC paths so there's some ugly hack involved that does the dirty work of remapping something into something else digestible by the WDS data-in operation.

Try to do a capture manually to see what speeds.