Hybrid Transactional/Analytical Processing (HTAP) on Hadoop
As businesses become more agile, the need for real-time and near real-time analysis on transactional data has become more important than ever. For database veterans, transactions and analytics have always been on two different systems. Such silo-ed approach resulted in expensive ETL processes, specialized data marts, SLA issues, and most importantly analytics on old data.
The latest architectural trends are increasingly leading the way to enable both transactions and analytics on the same data store. Gartner has called this capability of delivering transactions and analytics on the same data store and mixed workloads as Hybrid Transactional/Analytical Processing (HTAP).
Per Gartner, there are two types of HTAP – in-process HTAP and point-of-decision HTAP. The demand for HTAP has always existed from the business, but the technology limitations have forced IT not to deliver thus far.
- How can we change this scenario and help businesses become real-time?
- As the adoption of Hadoop and Big data continues, is there a way to leverage that infrastructure to achieve this database nirvana?
- Is it necessary to go all the way to in-memory computing (IMC) or can we leverage intermediate steps such as caching?