CapEx And OpEx Savings Are Nice!
But the value of NFV is service velocity and it all depends on hyperscale.
Written by Matt Harper
Today it is a lot simpler for network operators to deploy complex network functions as physical appliances than as virtualized, software solutions. The goal of ETSI’s Network Function Virtualization (NFV) working group and reference architecture is to change this. However, NFV won’t happen as currently envisioned unless there are compelling 10x economics driving actual deployments. To justify or make a business case, NFV must make networks 10X less expensive, 10X faster, 10X simpler, etc.
RIFT.io maintains close relationships with network operators worldwide; many of these operators are actively involved in NFV lab trials. In our discussions with these network operators, we ask them what they have discovered about NFV so far. What we have heard over and over is that the compelling thing NFV offers today is velocity of service deployment. Nobody has told us it costs less. Nobody has told us that it helps with OpEx. Nobody has told us they think it will reduce complexity. It’s much more than that — what NFV offers is faster service delivery and adaptability.
So, what is it going to take for network operators to achieve the radically reduced costs enjoyed by the largest cloud companies? In a word — Hyperscale.
Wikipedia contains a decent definition of hyperscale: https://en.wikipedia.org/wiki/Hyperscale
In computing, hyperscale is the ability of an architecture to scale appropriately as increased demand is added to the system. This typically involves the ability to seamlessly provision and add compute, memory, networking, and storage resources to a given node or set of nodes that make up a larger computing, distributed computing, or grid computing environment. Hyperscale computing is necessary in order to build a robust and scalable cloud, big data, map reduce, or distributed storage system and is often associated with the infrastructure required to run large distributed sites such as Facebook, Google, Microsoft, or Amazon.
For network operators, hyperscale NFV will accomplish four things at scale:
- Reduce explicit network complexity
- Enable rapid deployment of services on distributed infrastructure
- Automate distributed management
- Scale down OpEx
The diagram below shows where we think widely deployed NFV solutions are at on the evolutionary path today.
RIFT.ware leads the industry firmly beyond Maturity Model Level 2 with an advanced NFV Management and Orchestration (MANO) framework supporting multiple cloud technologies along with automation and visualization. In addition, RIFT.ware lays the foundation for reaching NFV Deployment Maturity Model Levels 3 and 4. However, these layers require a collaborative effort between network operators, infrastructure providers, and Virtual Network Function (VNF) builders. Network Operators need hyperscale VNFs.
So, what are some of the key characteristics often associated with a hyperscale VNF?
- Distributed Execution (Multi-host, multi VM/container)
- Dynamic horizontal auto-scaling support
- Single point of service management
- Single service address with elastic I/O
- Independently scalable service layers (e.g. APIs, data-plane, control-plane, database, etc.)
- Automatic Load balancing
- Distributed resiliency
- Network topology aware resource allocation
- SDN Integration
Fortunately, RIFT.ware also provides key building blocks to simplify the creation of hyperscale VNFs. We call this collection of software modules the RIFT.ware Hyperscale Engine.
RIFT.io’s Hyperscale Engine provides distributed processing capabilities in four key areas: Security, management, control plane, and data plane. VNF developers can pick and choose which modules to integrate, as they are completely optional. The RIFT.ware platform is designed for VNF developers to build true, hyperscale-capable VNFs and for network operators to more easily deploy, manage and scale them in production environments. As Rene Tio said smartly in his recent blog post, “NFV will only be successful long-term if VNFs and network operators fully embrace cloud principles and hyperscale architecture.”
RIFT.io is really looking forward to the rest of 2015 and 2016 as we get ready to deliver RIFT.ware 4.0 and bring true hyperscale capabilities and economics to NFV.
Originally published at riftio.com.