NFV deployment–important considerations for operators
After years of talk, demos and proof-of-concepts, Network Functions Virtualization (NFV) is a reality. Several telecom service providers have taken the step from trials to commercial installations of telecom clouds lately. Achieving a successful implementation is to a large degree dependent on the capabilities of the infrastructure layer, also called NFVi. Which are some important considerations for operators?
Top 3 reasons for adopting NFV
According to analyst firm TBR and their “NFV/SDN Customer Adoption Study H1 2017”, the top three reasons for adopting NFV are improved automation, operational savings and being able to introduce applications faster than what is possible with traditional physical network functions.
Getting this right including applications, orchestration, virtualization layer and hardware, is easier said than done. Several challenges must be addressed in NFV environments, which in most cases are multi-vendor based. Operational complexity, life-cycle management complexity and performance management complexity are some areas of concerns.
If you upgrade the virtual infrastructure manager (VIM) with a new OpenStack release, how will that impact the overall system performance? If you do changes in the software-defined networking (SDN) layer, will that result in any compatibility problems? Can you swap hardware and still maintain telco grade characteristics?
Deploying NFVi efficiently with pre-integrated solutions
If you would work with operations of a telecom cloud serving millions of users, you probably want to know whom to call to get support and advice if you run into issues. You certainly want to talk to someone who knows not only the hardware, hypervisor or orchestration, but someone who understands the big picture.
This is where successful deployments and operations start–who is responsible for the whole? To assume that great task, vendors must know their solutions in detail and the environment they operate in. By using a pre-integrated approach, operators get robust working NFV infrastructure providing low TCO and short TTM while being both flexible and open. And they get a partner who can take responsibility beyond one or a few products only.
Creating a pre-integrated solution
A pre-integrated solution has been created out of components, suited to work optimally as part of a system. Operators will know that all included components are compatible with each other, what the system performance is and how to do lifecycle management of individual products without leading to any unwelcome surprises. Since it is pre-integrated, the level of system integration can be limited to make it work within the customer environment which is a big advantage compared to the decoupled solutions used in the early days of NFV.
Which are some of the pre-integration activities? The most time is used to analyze and verify the optimal configurations of the components and work on how different use cases and features can co-exist without failing or interfering. A great effort is required to achieve more deterministic characteristics and high performance, e.g. using CPU pinning and SR-IOV. It is complex to make them work in conjunction with high availability and evacuation/migration scenarios. Pre-integration is also about hardening of open source to make it ready for telco use cases.
Software-defined infrastructure creates flexibility
Critical to the success of NFVi is how easily it adapts to changing business needs. One of the important contributors to flexibility is to software-define the underlying network and infrastructure.
A software-defined infrastructure (SDI) delivers business agility by dynamically allocating its compute, storage, and network resources to the services that need them. By provisioning workload resources only when they are needed, and releasing them when they are no longer needed, it reduces the need for over-provisioning. And, by using standard or open-source interfaces to interconnect all the layers in the stack, it allows a single management platform to provide not only insights into load balancing, usage, and traffic patterns, but automation to speed service delivery and reduce cost.
Being open and flexible are musts in NFVi since technology evolves rapidly and new applications imposes additional requirements on the infrastructure resulting in the need of various changes. The best way to achieve flexibility and openness is to follow ETSI NFV architectural principles and use open source technologies, e.g. OpenStack and OpenDaylight and open APIs and interfaces.
Ericsson is publishing a series of eBriefs where we share some of our insights related to NFV transformation. The fourth and the fifth eBriefs about "Consider a pre-integrated solution to reduce risk " and “Software defined everything” are now available.
Want to know more about how pre-integrated solutions reduce risk?
Want to know more about "software-defined everything" ?
Read all the e-Briefs on NFV deployment