Parallel replication across formats in SAP HANA for scaling out mixed OLTP/OLAP workloads

Abstract

Modern in-memory database systems are facing the need of efficiently supporting mixed workloads of OLTP and OLAP. A conventional approach to this requirement is to rely on ETL-style, application-driven data replication between two very different OLTP and OLAP systems, sacrificing realtime reporting on operational data. An alternative approach is to run OLTP and OLAP workloads in a single machine, which eventually limits the maximum scalability of OLAP query performance. In order to tackle this challenging problem, we propose a novel database replication architecture called Asynchronous Parallel Table Replication (ATR). ATR supports OLTP workloads in one primary machine, while it supports heavy OLAP workloads in replicas. Here, rowstore formats can be used for OLTP transactions at the primary, while column-store formats are used for OLAP analytical queries at the replicas. ATR is designed to support elastic scalability of OLAP query performance while it minimizes the overhead for transaction processing at the primary and minimizes CPU consumption for replayed transactions at the replicas. ATR employs a novel optimistic lock-free parallel log replay scheme which exploits characteristics of multi-version concurrency control (MVCC) in order to enable real-time reporting by minimizing the propagation delay between the primary and replicas. Through extensive experiments with a concrete implementation available in a commercial database system, we demonstrate that ATR achieves sub-second visibility delay even for updateintensive workloads, providing scalable OLAP performance without notable overhead to the primary.

Enabling High-Performance and Energy-Efficient Hybrid Transactional/Analytical Databases with Hardware/Software Cooperation

“…To enable real-time data analysis, where data analysis needs to be performed using the most recent version of data, a DBMS needs to be capable of efficiently executing analytics on fresh (i.e., the most recent) version of data that is ingested by transactional workloads, which is a challenging task. Several works from industry (e.g., [5,12,[118][119][120]) and academia (e.g., [17,18,115,[121][122][123][124][125][126][127] attempt to address issues with data freshness by proposing various techniques to support both transactional and analytical workloads in a single database system. This combined approach is known as hybrid transactional and analytical processing (HTAP).…”

Polynesia: Enabling Effective Hybrid Transactional/Analytical Databases with Specialized Hardware/Software Co-Design

“…To enable real-time analysis, we need a DBMS that allows us to run analytics on fresh data ingested by transactional engines. Several works from industry (e.g., [23,27,44,45,46]) and academia (e.g., [8,10,11,29,38,41,48,51,54,64] attempt to address issues with data freshness by proposing various techniques to support both transactional workloads and analytical workloads in a single database system. This combined approach is known as hybrid transactional and analytical processing (HTAP).…”

“…HTAP Systems. Several works from industry (e.g., [23,27,44,45,46]) and academia (e.g., [8,10,11,29,38,41,48,51,54,64]) propose various techniques to support HTAP. Many of them use a single-instance design [8,11,23,29,38,64], while others are multiple-instance [2, 27,51].…”

OLxPBench: Real-time, Semantically Consistent, and Domain-specific are Essential in Benchmarking, Designing, and Implementing HTAP Systems

“…In recent years, hybrid transaction/analytical processing (in short, HTAP) systems are proliferating fast. Many giant companies provide HTAP systems, including Oracle database in-memory [24], SQL Server [21], SAP HANA [34], Mem-SQL [14] and TiDB [4]. HTAP systems are popular for two reasons.…”

“…Due to several unfavorable factors such as strong consistency semantics and double operating cost, it is not easy to use separate stream processing systems as HTAP solutions. The other architecture uses a single HTAP DBMS [4], [25], [33], [34], achieving high performance for OLTP and OLAP. It eliminates data movement overhead but adds pressure to performance isolation or consistency guarantees.…”