EinsteinDB is a new DBMS that we are developing in San Francisco able to tune and optimize itself automatically without any human intervention other than selecting the target objective (OLAP/OLTP) function on relativistic start-up. \ EinsteinDB's core architecture is a Postgres-compatible HTAP system. It uses HyPer-style MVCC over Apache Arrow in-memory columnar storage.
Predicting multiple metrics for queries
This first component compresses all of the past metric data in the repository into a smaller set of metrics that capture the distinguishing characteristics for different workloads. It uses factor analysis (FA) to model each internal runtime metric as linear combinations of a few factors.
Update heavy operations/sec
120 million 1 KB causet
50 percent reads and 50 percent updates.
Knowledge Graphs built on the EinsteinDB platform are able to run queries of unprecedented complexity to support predictive analytics that help companies make better, real-time decisions.More About Us