Although moving to Windows Terminal Server and ACP Enabled Thin Clients will save a great deal of setup and maintenance, users have come to realize that moving to a centralized server concept also means increasing dependency on a single piece of equipment, the server. ACP recognizes the importance of developing a system that provides a very reliable platform for the hosting of user software, and so we have now addressed this issue by providing true Fail-Over redundancy. In a Thin Client system, clients can be replaced whenever needed without interrupting the application running on the client, but there needs to be some mechanism for handling the demise of one of the servers. Without this protection, a server failure can mean the loss of every client connected to that server.
While there are applications that can be purchased from other vendors (like Citrix) to provide for continued client operation even in the face of server failure, ACP wanted something that we could distribute as a part of our standard server package, ThinManager. A feature requested by many of our customers, we have been providing server protection as part of our standard product for several years now.
ACP Fail-Over / Redundancy is provided through an ongoing communication between all enabled Thin Clients and every Terminal Server on the network. This continuous communication allows the client to detect a problem with its current server and use its configuration settings to determine which of the still running Terminal Servers can host its session. The backup server is usually identified when the client is put into operation, but can be changed for individual clients (or groups of clients) at anytime.
The system administrator provides for the client a list of available servers. He can even designate a primary server, causing the client to connect to that client whenever it is available.
This server Fail-Over works not only in cases where there is a problem with the server itself, but also in situations where communication with the server has failed. In fact, when we demonstrate Fail-Over to customers we don't crash the server - we simply unplug it from the hub. This results in a momentary freezing of the client's screen, followed in a matter of seconds by the client coming back to life, but this time presenting a screen generated from the backup server.
When a Terminal Server becomes unavailable, the connected clients do not reboot. Instead, each client searches out and connects to the next listed Fail-Over Terminal Server. And if there is a session for this client already running on the newly located server, then it reconnects to the established session. When combined with dual-Ethernet Thin Clients such as the MBPC-5823 from Advantech, the system becomes even more secure by allowing the primary and backup servers to be on independent networks.
ACP has taken Fail-Over to a new level with our optional 'instant' Fail-Over. Under this model, each client is always running on two servers at the same time, but only displaying one of the sessions. In there are problems with the primary server, the client immediately switches to the display from the backup server. If the server sessions are presenting the same display, the user will probably not even notice the difference.
It is also possible to 'hotkey' between the two sessions, allowing the user to run two incompatible applications from the same Thin Client. A keystroke at the Thin Client's keyboard toggles to the second server and back again.
Many users also use the Fail-Over feature to install or update software. In a running system, all of the clients from a particular server can be switched over to a backup, allowing production to continue. Changes can be made to the primary server, and the clients can be added back a few at a time. If there is any problem with the new configuration, the clients can go right back to the backup server without affecting operation. When the update is complete, the other server(s) can be updated in a similar manner.
Duplicating the server reduces the chances of catastrophic failure with only a slight increase in the administration costs, and for many the security provided by a back-up server is worth the extra effort. If even higher server availability is required, contact ACP for information on connecting multiple servers together through hardware, providing for absolutely seamless transition from one server to the next.
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