Verify if ByteBuffer should be used everywhere

[jankotek]

With mapped ByteBuffer we could avoid some byte[] copying and improve performance. So we need to verify if using ByteBuffer all the way to DataInputOutput actually improves performance.

[jankotek]

No, it should not. Direct buffers are much slower for incremental writes than heap buffers. Bulk byte[] copy is fine. Rewriting core would not bring much performance gain