Present-day enterprises are increasingly turning to network virtualization as an alternative to physical devices. This is especially true in branches where there are often space, power, and support constraints. By embracing network function virtualization (NFV), it is possible to reduce branch network infrastructure to a single rack unit – a compute node on which virtual appliances like virtual firewalls, virtual SDWAN appliances, and virtual branch gateways can be deployed on-demand. Original equipment manufacturers (OEMs) are providing all-in-one solutions which include compute, management and orchestration tools, and virtual appliances for turnkey deployments. Alternatively, an enterprise can also choose a variety of commercially available virtualized network functions (VNFs) and off-the-shelf servers.
These are several questions an enterprise network administrator must ask the solution provider in order to choose the right NFV solution, including:
- How will I be alerted in case of any failures?
- What happens when my virtual appliance crashes? Can I recover its configuration? Will there be another appliance spun up in its place with minimal service disruption?
- What happens when capacity is exhausted on the server? Will my network be operational?
- How can I ensure that I can dynamically increase capacity while my service is up and running?
- How can I create a new service chain or delete an existing one?
- How can I secure my NFV deployment?
Choosing the right virtual appliances is critical to the success of any NFV deployment. Below are some important criteria for selecting the right virtual appliances for your environment:
- High available architecture (OEM should support a redundant design with 2 VNFs for HA)
- Standards-based data-representation of configuration (e.g. TOSCA/YAML/YANG)
- API support (preferably XML/JSON payloads)
- Horizontal scaling support (adding more VNFs in parallel for high throughput)
- Good logging and reconciliation mechanisms (failed config pushes, failed restart handling, etc.)
- Call home features
- Zero-touch provisioning features
- Support for multi-tenant environments
- Support for multiple hypervisors
- Support for containerization
- OEM ecosystem for VNF development (ability to test multiple VNFs with each other in POCs)
- Readily available cross VNF integrations (VNF to VNF communication) – for example, a virtual Firewall and virtual SDWAN VNF can be integrated to provide a branch SASE solution
Rahi is a leading global reseller and systems integrator. Contact us today to learn more about NFV solutions, best practices, and design considerations offered by leading OEMs.