Achieving high performance for 1 user is no use to anyone. We need to develop systems that scale to multi users. To achieve this you need to reduce contention. Unfortunately many batch processing scenarios are just that operations that run on a batch and as such impose locks of locks and seriously impact contention. The reason they operate on a batch is to obtain ultimate performance, doing an operation one by one just doesn’t scale. To reduce contention but still achieve high performance batch processes you need to do things in parallel. Service broker enables you to have your batch process work in parallel and too boot it's transactional, which means if everything falls over you can easily restart. In this session we will look at how you can turn your batch processes that used to block and cause contention into parallel operations that ensure that your users aren’t affected by the batch process running.