Join our Sr Distributed System Engineer, Kaveh Vahedipour, to learn more about ArangoDB Agency on September 19th, 2017 (6PM CEST/12PM ET/ 9AM PT) – View the Recording.
Distributed systems have become the standard topology on which modern appliances live. While the advantages of distributing workload for both performance as well as fault-tolerance are obvious, the runtime flexible configuration of such deployment becomes non-trivial.
ArangoDB clusters are no different in that regard. A potentially large database cluster’s configuration is manipulated at runtime by addition, alteration and removal of collections, indexes, and even servers. All servers need to trust in a fault-tolerant centralized configuration tree, which we call “the agency” in arango-speak.
The agency is in effect a cluster within the cluster with very special qualities. It is a RAFT-based store for objects and services, which not only hold the larger cluster’s animal brain but also handles server monitoring and failover.
- offers a brief introduction into the RAFT consensus protocol and specifics of its implementation in ArangoDB
- discusses and shows how users may use such ArangoDB agency for their own service deployments
- enables one to administer the ArangoDB cluster’s agency
- illuminates aspects of performance and issues related, which need to be considered
September 19th, 2017 (6PM CEST/12PM ET/ 9AM PT)