Follow Us.
Become a Fan! 
One of the early realizations users have when they upgrade to Thin Clients is that the server is key. And there is no denying it - a Terminal Server requires a little special attention. Fortunately there are many very easy steps a system administrator can take to make sure there are no problems with the operation of the connected Thin Clients.
Protect the server just like you would protect any important PC. The most obvious thing to watch is the power, and so a high-quality Uninterruptible Power Supply is a must for any Window Terminal Server. This will protect data from power loss by not only providing constant power but eliminating brownouts and protecting the hardware from power spikes.
Once the machine has reliable power, the next thing to consider is protection of the data itself. A common solution for a server is to use a Redundant Arrays of Inexpensive Disks (RAID). There are various types of disk arrays identified by appending a number to the RAID acronym. The most common are RAID-0, RAID-1, RAID-3 and RAID-5. The basic idea of RAID is to combine multiple small, inexpensive disk drives into an array of disk drives to yield performance exceeding that of a single drive. Additionally, this array of drives appears to the computer as a single logical storage unit or drive. Several levels of RAID also allow for disks to be swapped while the machine is running (known as hot-swappable drives) that lets a user replace a damaged disk without bringing the server down.
A simple, inexpensive data protection scheme is the old standby - Tape Backup. It is a solution that is cheaper than a hot-swappable RAID system, and has the added benefit of providing for off-site storage. Having the server(s) in a single secure area greatly simplifies the process of producing and keeping good backups. The IT department is already backing up its other machines, and while they wouldn't want to wander around the plant and back up all the PCs in a factory, adding a few more servers is usually a very simple matter. And while it goes without saying that data recovery is much easier with current backups, it is also easier to restore an application if it has been kept current. Thin Clients make it easier to keep all your apps up to date, as the updates only have to take place on the servers.
Another issue that must be covered when a Terminal Server is installed is its security. Pick a good virus scan software and keep it updated. Also use some of the standard Windows, Citrix or ACP features (User Profiles, Initial applications, Published applications, etc.) to keep operators from having access to the Windows operating system. And don't overlook simple system maintenance. Sure you should implement an effective backup program, but you should also monitor disk usage, be aware of CPU utilization, and pay attention to the overall system architecture (failover strategy, network layout, etc.) as the number of clients grows.
A very popular enhancement that ACP has made to basic Terminal Server operation is the addition of ACP's Thin Client Failover feature. This is a truly automatic server failover strategy that is not limited to single failover server. Each client is configured with the IP addresses of backup servers. We even support something that we call "Enforced Primary", which means that when the client detects that its primary server has come back online it can be configured to switch back. An important distinction between ACP failover and other solutions is that all server switching is controlled by the clients, meaning that the client does not have to be rebooted when a server fails. This is essential in many plants as there is not ready access to the clients.
While the first systems installed to take advantage of ACP's Failover ability only used two servers (a primary and a backup), more and more systems are being installed with multiple servers. This allows many clients to be assigned so they distribute the load if their primary server has a problem. Failover is not only used to provide a backup in case of server failure - many customers use it to update software on the primary server. When the primary is brought down for maintenance or updates, all the clients switch to their backup servers. The new configuration can then be tested, and when it is ready and brought back online, all of it clients come back.
For some versions of Windows 2000 Server Microsoft also offers something they call a Cluster Server. It combines two machines to provide high-availability, but it does not scale the system performance. Two machines are linked with one external SCSI disk drive, and in the case of a server failure it will automatically switch the resources from the failed server to the good server, operating off of the single shared disk. This solution will not keep programs synchronized, and so for applications that keep data in memory the task will loose its context. This is not a problem for applications that are not dependant on non-volatile storage (like many database applications).
Microsoft also offers something called Windows Load Balancing Service, which will support up to 32 servers (or 'nodes') running together. It is TCP/IP based, and each cluster has an IP address assigned. This solution does scales the systems performance and adds fault tolerance at the same time.
Citrix Load Balancing requires Citrix Metaframe to be installed, but allows users to group multiple servers into server farms and routes users to the least busy server. Although Metaframe may be expensive for many locations, it is an excellent strategy when the installation becomes very large.
For completely reliable system operation, you need a server that will continue to run no matter what happens. Several companies provide approaches to this type of availability.
Marathon Technologies has a solution that uses 4 separate PCs, where the compute elements run applications in lock step and the I/O processors provide RAID-1 storage. Any component can fail without affecting the operation of the applications.
Newer to the PC world is a company called Stratus, which has been making high availability machines for years and now offers a very cost effective solution with their Windows 2000 based ftServer. This is fully replicated, fault-tolerant hardware where duplicate components operate in lockstep. It has hot-pluggable components, and Stratus has made modifications to Windows 2000 to enhance the reliability of the Operating System as well. Additionally, the system constantly monitors its operation and reports any exception with a phone call to support.
For more information on ACP Industrial Thin Client computers, please visit our web site at http://www.thinmanager.com
To sign up for the E-mail newsletter go here: ACP newsletter signup
For an archive of past newsletter articles go to: ACP Newsletter Archive
Copyright © Automation Control Products, Inc. All rights reserved.
