Abstrakt
One possible technique of data processing is its transformation into a data stream and the performance of particular operations on the data tuples. These operations can be usually processed concurrently especially when the plan of operations is branched.
Therefore, this way of data processing is suitable for evaluation in parallel environment. On the other hand, the ordering of the performance of tasks associated with the operations is closely related to the utilization of the hardware components.
For that reason, the task scheduler in these systems influences the efficiency a great deal. In this paper, we introduce our implementation of a task scheduler which deals well with various hardware factors such as memory caches and Non-Uniform Memory Access factor.