Ok a much improved version of my code. Now uses a custom logging class that asynchronously flushes a queue to the disk. Right now i have it set to flush every 100 readings but that can be changed in source code. I think this is the best we can do until they add an streamwriter to KSP.io without resorting to workarounds of the security model. Ive tried this out to 100MB of data in csv and on my computer i couldn\'t notice any slow down. edit: added a deconstructor so it should flush the log to disk on close as well.