White Papers

Understanding OPNFV

Issue link: https://read.uberflip.com/i/1344850

Contents of this Issue


Page 38 of 144

Understanding OPNFV 38 faster software development velocity and the ability to standardize on persistent database or storage layers. The database or storage layers could span the entire range of performance from memory, flash, block storage to object storage. ● Scale-out vs. scale-up: Cloud native apps scale by increasing the number of instances rather than adding more resources to a given instance. This is critical in terms of achieving hardware independence and performance that scales-out or in, in response to utilization. A corollary to scale-out is self-healing, where the system responds by creating new instances of a microservice if existing instances become unavailable. ● Antifragility: This is probably the most important yet neglected factor. An antifragile system is better than stable. It is a system that improves in stability as stress is applied to it in the form of infrastructure and other failures. Networking services have been designed to be antifragile since the Internet was first invented, but this is new for VNFs. In fact, Netflix has pioneered a new field of "chaos engineering" where they fail instances, networks, availability zones and entire regions to test to ensure antifragility. In production. All the time. In reality, it takes a long time to get to all of an organization's workloads to be cloud native. It took Netflix 7 years to completely move from traditional applications to cloud native! For most users, in the interim, many of the benefits of NFV can be realized even with cloud hosted VNFs (traditional VNFs virtualized) or cloud optimized VNFs (partially cloud native VNFs). In fact, most VNF vendors are taking this exact approach by simply virtualizing their physical network functions and doing some basic performance optimizations to begin with. Progression to Cloud Native VNFs

Articles in this issue

view archives of White Papers - Understanding OPNFV