A major constraint of deciding to use a Client model over individual developer databases (IBM i Slave), apart from the pros and cons listed, is the hardware specification of the client PCs and the server PC. If the server is relatively low specification, Network Client performance may be severely affected. On the other hand, if a powerful server is available the Network Client model becomes feasible.
A further difference with IBM i Slave is that the repository is shared so other developer's changes are immediately impacting the work environment - either positively as the updates are needed or negatively because they are defective. There is more control with an IBM i Slave, and even more with an Independent Windows Workstation to a VCS Master.
Note: Client Systems are sharing the database. This reduces the effectiveness of the redundancy afforded by having the Master System. If a PC database is lost, then the changes of ALL developers using that database up to the last check in will be lost. So using a Client model increases the need for a good PC database backup strategy to be implemented.