Source: Manuchi

A service mesh for collaboration between geo-distributed services: the replication case


Marie Delavergne, Ronan-Alexandre Cherrueau, and Adrien Lebre

LS2N, INRIA, France


Edge computing is becoming more and more present, with edge infrastructures geo-distributed around the globe. These infrastructures must be able to manage the latency and disconnections inherent to their distribution. In parallel, the behavior of cloud applications is getting more complex to apprehend as they are composed of a lot of small services, which serve each a defined purpose. To manage these applications more easily, service meshes handle communications between services and offer different features such as service discovery or traffic monitoring.

We believe that service meshes are an opportunity to create a service dedicated to manage the geo-distribution of applications. To do so, we created scope-lang, a language that allows DevOps to specify precisely where a request will be executed. By reifying the location of application instances in requests and allowing redirections between these requests thanks our service and some service mesh functionalities, we make possible three types of collaborations between services. This paper focuses on the replication of a resource on different services instances across different sites and presents Cheops, the service to manage geo-distribution.

Versions of the article


Link to the article (Version 0)

Previous versions