I have never ran out of memory with 1.5gb of heap so not too sure why you need more?
Moreover adding too much ram would actually slow the client down as java doesn't always handle large heaps well unless your machine has a lot of cores.
Neither do I run out of memory, what I want to achieve is just speed up my analysis.
Currently I allocated it with 14336MB, still sometimes analysis can be slow. (Well, most of the time I do analysis with past 2 or 3 hours data with unlimited purepath)
Anyway, my laptop has 4 CPU core and 16GB RAM.
I will warn you also, that the way that Java memory works is that the heap is just one component of it. the extra ~2 GB RAM you have will not support the OS + the native process memory requirements for the client, let alone the client launcher process and any other processes you may have running concurrently on your machine. The most I have ever set my client to is 4 GB, though I have almost always had a 16 GB workstation
adding RAM won't speed up your analysis. You should be creating measures on the key elements you are monitoring as they are very quick and work on long time periods unlike the purepaths.
To be honest, you are probably putting of a lot of stress on the front-end server by doing this.
The tool is not meant to be used to systematically analyse hours of purepaths as it is too time consuming unless you have very little traffic. I practically never look at more than 30 minutes of data when troubleshooting. On busy environments even 5 minutes can produce hundred of thousands of purepaths.
Without charts and measures you won't be able to see the wood from the trees. It is impossible to spot a slow down for instance just by looking at all your purepaths however a response time chart would show it immediately.
They are plenty of good materials on APM university (https://university.dynatrace.com/home) which explain how the tool is best used.