Red Hat ブログ
Network functions virtualization (NFV) deployments continue to grow in the telco sector. Top service providers from all over the world are moving to virtualize 60 to 80 percent of their networking functions in the next three to five years, including AT&T, China Telecom and Vodafone Germany, according to research from Light Reading. If you’re ready to get going on your NFV deployment, we have seven “must have” tips for an effective telco NFV deployment to help you get started, which Red Hat and Juniper Networks shared at the OpenStack Summit last month.
Presenters Anita Tragler, Red Hat’s product manager for networking and NFV, and Greg Smith, product manager in Juniper Networks’ broadband group, started off with an explanation of how NFV works. NFV extracts network functions like firewalls and routers from traditional network appliances and runs them as software on commercial, off-the-shelf servers.
Then the duo reviewed seven tough and typical challenges faced with NFV deployments. They looked at a virtual broadband network deployment as the specific application, with the caveat that “if you had to do it today, or next week, how would you do it,” Smith said.
Here are the seven “must haves” for a highly effective telco NFV deployment:
- Predictable high performance
- Superior layer 2 and layer 3 connectivity
- Efficient multi-site deployments
- A programmable infrastructure
- High availability
- Distributed security
Tragler and Smith detailed each of these challenge areas and then explained how OpenStack can help solve them. For telcos, predictable high performance can’t be sacrificed. OpenStack delivers through a combination of data-path options and pformance tuning. Among the capabilities to deliver superior layer 2 and layer 3 connectivity is a new feature in OpenStack Neutron, a software-defined network (SDN) project focused on delivering networking-as-a-service (NaaS) in virtual compute environments.
Support in OpenStack for multi-site deployments that deliver better scale, redundancy and disaster recovery are currently a big topic of discussion in the OpenStack community. New community projects are emerging, and according to Tragler there are two options available now designed to improve multi-site deployments.
For NFV to work for carriers delivering services over virtual broadband networks, it also has to support multi-tenancy, essentially different kinds of virtual network functions (VNFs) which each have their own path through the cloud. According to Smith, there are “many different kinds of paths through the network and many different kinds of applications on compute nodes.” Work is already underway, via manageable, scalable overlay tunnels, to address these challenges, Tragler told attendees.
Programmability is critical too, so providers can more easily deliver value added services. What’s wanted is “at the click of a button, and that’s the hard part, I think, to be able to chain these services on demand,” said Tragler. For that, OpenStack and Neutron need to integrate with the SDN controller. And there’s work in progress for adding support for a new dynamic header that allows multiple layers of service, she said. High availability and distributed security are also key, and Tragler and Smith also discuss solutions for these two must-haves .
The session provided real-world advice for attendees to take home and put to use in their own NFV deployments, and provided a planning framework for those exploring NFV. Tragler also provided a brief overview of Red Hat products, including Red Hat OpenStack Platform and Red Hat Enterprise Linux, and how they fit in an NFV environment.
To get all of the technical details and implementation tips for these seven areas, check out the video of the full presentation.