Sandeep,
When you call a triggered task, there is some overhead in initiating the pipeline, executing the pipeline, tearing down the pipeline. That overheard of initiating and closing is removed by using Ultra Pipelines. An Ultra Pipeline will start once then remain running, listening for new calls. In the even of a failure, there is a new type of node called a Feedmaster that is always watching an will attempt to start the pipeline again so it’s ready for the next call. You can have multiple executions of a pipeline running on multiple nodes so you can tune it to meet your throughput needs.