BABEL

The project BABEL was a 2019 winner of the PREMATURATION call for projects, led by the Innovation Lab of IP Paris.

Reliable and scalable distributed storage systems should handle efficiently where the data should be stored to minimize both the costs and access latency. In addition, they should provide proper mechanisms for data availability in case of node failures. However, traditional data placement and data availability strategies prove to be costly to properly scale up and provide better data availability.

Babel is a set of software solutions for distributed storage systems that aims reducing storage systems cost and improving user quality of experience. It integrates a solution for data placement using machine learning to predict the suitable data storage placement for each type of information, reducing thus the cost of relocating and optimizing response rate. In addition, Babel deploys an innovative erasure coding technique that optimizes repair bandwidth, I/O and required recovery time.

• Flexible AI-powered automated data placement for data storage systems. Using online machine learning to optimize data placement.

• Efficient erasure coding to improve data availability and failure repair. 

• On-the-fly and multi-features data placement

• Scalable and reliable data storage

• Reduced infrastructure cost

• Better quality of experience

Babel allows cost savings, via:

• Optimized usage of expensive hard drives

• Reduced network utilization for data relocation

• Reduced data access latency

• Reduced repair bandwidth and repair I/O

• Decreased recovery time