Google has published a paper about Mesa, its new data warehousing system, used specifically to handle its online advertising business data. It’s a system used to create reports and analyzes for the company mangers. Due to the enormous amounts of data managed globally by Google, it’s a very complex system, requiring something more than other systems created by Google over the years.
Mesa’s possibilities are really impressive and give an idea of what is the amount of data that Google manages with AdWords / AdSense. The system works across multiple data centers and we know that a Google data center is impressive because some time ago the company showed one to the public. Even if one of the data centers that are part of the system went out of service for some reason or was deactivated, Mesa would keep on working with the others.
Mesa is made to handle petabytes of data, the equivalent of thousands of hard disks of the models currently used in desktop computers. Every day it manages billions of queries that read trillions of data inside it. The replication of data across various data centers allows to get with consistent and repeatable answers to queries with low latency.
To do this, the Mesa database must comply to the ACID (Atomicity, Consistency, Isolation, and Durability) properties. The properties whose names form the acronym must be followed to ensure from the technical point of view the reliability of the system.
Google will present the Mesa system to the Conference on Very Large Data Bases, to be held next month in Hangzhou, China. There is curiosity about this technology also thinking about its future developments and applications. The paper about Mesa mentions the usefulness of the paradigm of cloud computing in conjunction with a decentralized architecture. Not many companies need systems of that type but Google could develop Mesa to use it beyond data warehousing. Sometimes Google’s plans are long term so it may take some time to understand the evolution of this system.
[ad name=”AmazonElectronics468″]
