Apr 2019 – Feb 2020 · 10 mos · Bengaluru Area, India

• Worked as a part of HIVE team. Focused on projects to make queries interactive and faster.

Jul 2011 – Apr 2019 · 7 yrs 9 mos · Pune Area, India

• Unbalanced Partition in HANA
• Introducing a new flexible multilevel partitioning scheme in HANA. Replacing traditional scheme where user is limited to two level partitioning and fixed format. Enables user to create a suitable partitioning structure based on data distribution, skewness as well as use of a new extended property set.
• Synopsis Partition Pruning in HANA
• Extends scope of partition pruning beyond partition column. Metadata for these columns are dynamically maintained and used during query execution to prune an irrelevant partition.
• Simplified Native Access Plan
• Achieves fast query processing by just-in-time compilation of incoming queries. By producing machine code at run time we avoid the overhead of traditional interpretation systems. Generated code is tailor made for each individual query and this results in reduced number of machine instruction and better instruction locality.
• Data Federation from ASE to Other Remote Server
• Adds ability in ASE to talk to non-ASE remote server and execute a query which may use table from both local and remote server. ASE prepares query plan and depending on the plan, query could be fully or partially executed at remote server.
• Precompiled Result Set
• PRS is equivalent to materialized view. Result set of a query is materialized and maintained when the base tables change. The main challenge is to automatically refresh the resultset when base tables change without recomputing the whole query. To optimize, we just compute the delta and apply in the resultset.