Omer is currently a software developer for Huawei, and a core contributor for Dragonflow in OpenStack. He has 9 years of professional programming experience, with over 5 years expertise in Linux systems and networking. Omer has a B.Sc in physics, and is working towards his M.Sc. in Computer Science.
In large-scale OpenStack clouds (more than 1,000 nodes), OpenStack database layer cannot scale-out and a new paradigm of persistent data mechanism should be re-defined for OpenStack. In SDN world, most of the current solutions for OpenStack rely on open source NoSQL servers to store persistent network topologies and objects and some of them are successfully working in production. However, due to the fact that database backends vary on consistency, availability and network partition, we designed a general database consistency solution, based on distributed lock manager, object versioning and high-performance messaging layer to synchronize the network topologies and objects among different database backends. Finally we achieves the desired consistency for the whole Dragonflow platform.
