Oh, just if you are interested in more detailed information:
If to speak about performance only, not taking quality into account at all, you may wonder when JpegWriter will be faster.
It will be in case when you load image one time and then save it multiple times. Therefore your application will perform following actions: load, rotate, save, save, save, save.
While LosslessJpegTransform always reads source file and then writes the result, so it will be: load, saveRotated, load, saveRotated, load, saveRotated, load, saveRotated - 8 actions instead of 6 in the example above.
But (again! ;)) - it is rare case when you will just save the same image multiple times. More common scenario is something like that: load, flip, save, crop, save, rotate, save, anotherFlip, save - 9 actions. LosslessJpegTransform will perform 8 actions to get the same result.
And - not the only action count is important. JpegTransform always has to read/write data from/to the disk, it is much slower than operations on Bitmaps in memory. But from the other hand, LosslessJpegTransform doesn't actually decode JPEG blocks - so it avoids a lot of calculations in his work.
The idea is that you always should profile your application in your enviroment before optimization. In my tests I have got different results - depending on the source image size and set of used transforms.
Our team is always ready to answer your questions, and the more information you provide, the better answers we give ;) Otherwise we are able to give only most common advices.