At this moment, compression tasks are managed by the means of logback-core, for example, TimeBasedRollingPolicy creates AsynCompressor task for each policy instance and creates as many compression threads, as many TimeBasedRollingPolicy instances present in the logger context. In the systems with great number of "fat" log files it may cause performance issues with filesystem.
It would be much better if logback-core allowed to define external thread pool executor service and reuse it for such async tasks.

Thanks