Well the main thing to remember when evaluating the MSM is that it is a useful basis but a vast over simplification. Whilst there is plenty of evidence for separate memory stores for long and short term memories, there is also evidence for these long term and short term memories being subdivided into different memory stores. If it helps, you can imagine the long and short term stores as two different houses, but within these houses are different rooms in which different types of memories are stored. In your exam, you'll want to discuss the fact that there is good evidence that the MSM is correct when it identifies the long and short term memory stores as separate.
There are two case studies (amongst other evidence) that are easy to remember and useful for demonstrating the strength of the MSM. Firstly, there is a case study of a man known as KF who was involved in a motorbike crash. After the crash, this man's short term memory was severely damaged, but his long term memory was in tact. This suggests that there are separate places in the brain where the long and short term memories are stored, because otherwise both types of memory would have been damaged at the same time. The fact that only one was damaged suggests that the different types of memory are stored in different parts of the brain. Another case study, concerning a man called HM who went through brain surgery, shows something similar. However, whilst KF's short term memory was damaged and long term memory was fine, HM's memory was the other way around - he could retain information in the short term but forgot everything in the long term. This adds concurrent validity to the theory that the long and short term memory stores are separated. In this context, concurrent validity means that the theory that the stores are separate has successfully predicted that both the long term memory could stay in tact whilst the short term was damaged, and vice versa.