# `HashAgg` on TiKV Servers as a Parallel Task

![`smart`](HASHAGG08_aa_003_20231013.png)

------------------------------------------------------------------------------
Additional optimization can be achieved through parallelism. For example, the original table can be divided into several segments for concurrent scanning, as illustrated in the slides.

In this case, the table is split into two parts, allowing for parallel scanning. This approach generates two temporary result tables, both stored in cache, as shown in the slide.