- Peloton is a self-driving in-memory relational DBMS for real-time analytics.
- It contains domain-specific AI for automatically adapting to evolving real-world workloads.
- It is designed from the ground up to leverage the characteristics of fast non-volatile memory technologies.
- It can handle both fast ACID transactions and complex analytical queries on the same database.
The current trend is to manually tune the DBMS configuration for evolving real-world workloads. This approach requires the database administrator to constantly adapt the DBMS based on the current query workload. The adminstrator needs to understand the subtle interactions between the different knobs exposed by the system to do this kind of black-box tuning. Further, it is often the case that several critical parameters used within the DBMS are not exposed as knobs to the administrator.
Peloton is designed to automate some of the critical tasks performed by the database administrator. Using novel physical design algorithms and domain-specific AI, it can automatically and incrementally adapt the storage layout, access methods, and data placement policy employed inside the DBMS in tandem with workload shifts.
For more details, please visit the Peloton Wiki page.
Check out the installation instructions.
We invite you to help us build the future of self-driving DBMSs. Please look up the contributing guide for details.
Before reporting a problem, check out this how to file an issue guide.
Technology preview: currently unsupported, may be functionally incomplete or unsuitable for production use.
See the contributors page.
Copyright (c) 2015-16 CMU Database Group
Licensed under the Apache License.