In less than 12 months Windows Server 2003 will be put to rest like its desktop sibling Windows XP. With this end of life date looming I have been working to migrate my clients to the latest version of Windows Server. This process is sometimes difficult to get started because I find many people don’t know or understand what their server actually does. This article is going to serve as an overview, a sort of crash course in what roles your server plays on your network. With Windows Servers there are three main roles I am going to discuss for simplicity sake. First is the Domain Controller, second is the File/Print Server, and third is the Application Server.
The domain controller server roll is the most complicated and least forgiving role when upgrading and migrating. Domain controllers keep a list of user names and passwords, they also keep track of what security rights and privileges users have on their computers and on the servers, kind of like “big brother” for your network. Anytime a user logs in the domain controller is consulted to see if they are allowed. When a user opens up a file on that “x” drive everyone has, the domain controller is asked if the user is allowed to access that file. In most cases the domain controller also provides some other services it needs but also shares with the rest of the devices on the network. One important service is called DNS or the Domain Naming Service. If you have ever typed in a web address or had to browse to a device on your network you have made use of DNS. It is like an address directory for the network it translates computer names like “Jim’s Computer” to a numerical address on the network. The other service usually found on Domain controllers in smaller networks is DHCP. DHCP is the other side of the coin from DNS. When your computer starts up it asks the DHCP service for an address so other computers can find it on the network. The best practice is to let the domain controller do its job and not use it for any other roles. This rarely happens in smaller networks though generally due to initial cost as well as additional maintenance cost for having multiple servers.
Our second role is that of the file and print server. These are two distinct roles but generally are lumped together except in larger networks where usage demands they be split. The file server is really where all the files are physically stored that are on that mysterious “x” drive I mentioned. By the way your, “x” drive might be called p, or z, or t, or any other letter of the alphabet depending on how your IT Pro felt when they set it up. The sole purpose of the file server portion of our dual identity server role is just to store your files and wait diligently until you need to retrieve your files from it. The print server identity is similar in simplistic duty. It waits for you to send a document to print, gathers it all up, and forwards it on to the printer. Additionally, the print server makes it easier on us administrators to manage the printers shared on the network and the drivers needed to install them on all the computers on your network.
Last but not least is the application server. These servers run specific programs like databases like SQL or Oracle, or email servers like Exchange. Sometimes they run multiple applications and perform the duties of a file server. The application server as its most basic function runs an application and stores all the data that application uses or creates. Turn it off and your network accounting software doesn’t work or outlook can’t send or receive email any more.
In small businesses all of these roles are usually stuffed into one single box which is a very cost effective use of the hardware. The upside of having a single server is maintaining it is less expensive because there is only one piece of hardware to worry about and only one device to backup. The downside is this set up isn’t very flexible and can be difficult and time consuming to migrate when it is time to replace or upgrade the server. Migration is especially tricky with the domain controller role running on that single server as well. Splitting up the roles on different hardware makes the system more flexible and resilient. If one server goes down you are not completely dead in the water. You still have access to the remaining services. Many times the domain controller role is set up on at least two servers in order to insure basic network services are always available. Multiple servers though means more cost and more hardware to manage not to mention multiple backups with which to deal. Scaling the hardware in each server can help to mitigate the cost somewhat by choosing hardware that has just the processing power, RAM, and storage needed to perform the role it is designed for.
I hope this brief overview helps you better understand what your servers do and some of the benefits and drawbacks to having a single server or multiple servers on your network. I recommend you have a plan and start with a single server that does it all. Then in a year or two add a separate domain controller and remove the role from that first server. If your applications are running slow or you are running out of space add a server just for the applications and alleviate that first server of those duties as well. Soon it will be time to replace that old server and you will have a solid, reliable, scalable server infrastructure that won’t even blink under the pressures of your business.