We have people who work on half a dozen projects at a time (tech writers, testers, etc.). What's the best way to do time planning for resources like this if you have six separate MS Project Plans for the various projects they are working on? I have heard MS Project Server solves this problem, but is it worth the grief?
Are the projects in any way linked or dependent upon each other, other than shared resources, e.g., is an output of one project an input to another? Also, I assume these projects start and stop in a seemingly random way, not coordinated with each other, right?
If both of these questions are true, then it sounds like the issue you need solve is to ensure your human resources are and remain leveled at the utilization level at which they were hired, e.g., 50%, 100%.
Leveling is important but it can become a waste of time quickly. As a project, or projects, progress, things change and your resources are no longer leveled. To chase that is to chase your tail.
I think the same phenomena would apply here, especially if these projects are not in any way integrated. The effort to pull them into an integrated master schedule (IMS) would quickly exceed any benefit you get.
Instead, I would maintain a separate human resource calendar, in which I would document who is assigned where and at what utilization level (planned, not actual). For a FTE, I would assume maybe 75%, maybe 80%, leaving 20% to 25% of non productive, normal activities in an employee's life to include vacations, training, uncertain sick leave, education, etc. It also can be a bit of contingency if a project finds they need to crash a bit and increase an employee's op tempo for a time being.
Then when a project is being scheduled, they will estimate and plan based on what utilization level they have for a specific labor category. If you have 15% remaining for your tech writer, than that is what you use in your time estimate to arrive at a duration.
I wrote above planned, not actual, because a plan value is the independent variable; you control it. The actual variable is dependent; you measure it. If variances are starting to accrue unfavorably including your over time expenses, i.e., your utilization estimates were too optimistic, then you can intervene and make the tough decisions to slow a project down or hire.