It is expected that at any given time during a normal day, there will be between 50 and 100 members of the public interacting with each individual server in the cluster. If the network card in one of the servers breaks while being used, it is very likely that the users connected to it would still be able to complete their current operations without any inconvenience.

Explain why this hardware problem would not seriously affect the users who are currently connected to that particular server.

The answer could include the following:

Redundancy:

• There is another network card in the server which is waiting to take the role of the first.
• The user session information is replicated onto another server so that the other server can take over the operation of that client, should the first one be completely unavailable (e.g. If there is no redundant network card).

Failover:

• At the point failure, the OS realises that the network card is not working and automatically switches to the redundant network card to continue conversation with the client.
• At the point of failure, the load balancer/application server/partner server realises that the server is no longer reachable and from that point onwards all requests by the owner of that session are directed to the server which contains a replicated copy of the original session. The session is then replicated to a third machine.

Replication of session information:

• The information contained in the session of a user can either be stored only in one node of the cluster (sticky sessions), or it can be stored in shared memory or secondary storage accessible by the whole cluster. There are performance overheads, because if the session information is constantly changing, or it has to be accessed with each request then it would usually be much faster to keep it in primary memory in one server (sticky sessions) to which the user is always directed by the load balancer / application server.

[1–2 marks]
A limited response that indicates very little understanding of the topic or the reason is not clear. Uses little or no appropriate subject specific terminology. No reference is made to the scenario in the stimulus material. The response is theoretical and descriptive.

[3–4 marks]
A superficial explanation of why this hardware failure would not seriously affect the users who are currently connected to that particular server. There is some use of appropriate subject specific terminology in the response.

[5–6 marks]
A thorough explanation of why this hardware failure would not seriously affect the users who are currently connected to that particular server. Explicit and relevant references are made to the scenario in the stimulus material. There is appropriate use of subject specific terminology throughout the response.