A brief introduction to Dbseer algorithm

Researchers at MIT have developed an algorithm called Dbseer, according to Gigaom's Jordan Novet. The algorithm can improve the performance of cloud database and reduce the user's demand for cost.

Here's what I did:

The algorithm above is designed to make the cloud infrastructure of the database more efficient and to drive more things like workloads on fewer servers than to distribute them as widely as possible.

Many database companies are now responsible for allocating responsibilities to maintain a lower latency. But from the cloud provider's point of view, using the researcher's Dbseer algorithm, you can improve the performance of the cloud database.

A recent paper published by the researchers said that infrastructure as a service (IaaS) provider running virtual machines on the server may not be the most efficient method for databases because resources cannot be shared between applications running on any given server. This paper better observes the current workload, predicts future workloads, and brings together different types of load on different servers. The researchers point out that cloud providers can adjust service level agreements to commit to a certain amount of latency without requiring customers to pay for the number and size of virtual machines.

Dbseer may also be of interest to database applications and server vendors, and Teradata, a global enterprise Data Warehouse solution leader, is interested in proprietary software for the algorithm. At the same time, he is now working at Microsoft's Carlo Curino, one of MIT's researchers. In addition, the Chinese network server provider Quanta has also provided financial support for the study.

So far, on GitHub, Dbseer has only proven to be able to accurately predict workload requirements for transactional MySQL databases. Further research will be necessary to apply the algorithm to other database management systems.

Changing thinking can produce good economic meaning. In the cloud provider's data center, the more hardware customers use, the more expensive the price. If these devices can work more efficiently, the cost of users may be reduced.