T 4.82 Failed or unavailable terminal servers

In a classic client/server architecture, applications such as text processing are executed on the clients of the users. If there is currently no contact to a required server, it is often possible to continue to work on the clients asynchronously depending on the application. For example, an open document may also be stored to the file server at a later point in time if the network has failed. If an email server is unavailable for a short period, the message is usually buffered and transmitted in a delayed manner by the email client. It is possible that the user does not even recognise such an incident.

In a terminal server environment however, the applications are executed centrally and their output is transmitted to the corresponding terminal. If the terminal server is unavailable, no user input can be processed and the applications provided by the server will fail immediately. If the clients, e.g. as is usual in a thin-client environment, obtain the entire user interface, including the operating system, from the remote server, the IT system will fail completely from the user's point of view.

Normally, only individual users are affected by failures of the network or the terminal server. In many cases, numerous or even all clients of the organisation rely upon the terminal server. If a terminal server fails, a large number of users will be affected simultaneously in this case.

Example:

A municipal service provider introduces terminal servers for several of its business units. For this, several terminal servers are procured, but every business unit is provided with its own terminal server instead of a centralised terminal server farm. This way, the expenses for load balancers are not applicable and the communication paths between client and server can be shortened.

After the terminal servers worked in a stable manner for a couple of months, one server fails and the entire business unit is inoperable until the repair is performed. After a few more months, another business unit expanded and was able to hire numerous additional employees. The increased utilisation of the business unit's terminal server related to this was compensated by distributing the new employees to the less utilised terminal servers of the other business units. This results in unstructured "chaos" on the terminal servers and in significantly increased organisational efforts.