CSP Evolution

An online magazine tracking the transformation of the 21st century digital communications service provider

Welcome to CSP Evolution, a quarterly online publication tracking the far-reaching changes currently sweeping across what we used to call the communications service provider/telecom industry.

Facing challenges on multiple fronts – exploding network requirements, increasing customer expectations, and a changing competitive landscape – CSPs recognise they have the opportunity to leverage their strengths and become purveyors of ‘digital services’ by optimizing current connectivity-based offers to improve service and costs and creating value from new, innovative operating models and service offers. This will require a raft of technical and cultural changes and an overall infrastructure and operational transformation for them to prosper in a hyper-competitive environment.

Arguably the greatest competition facing CSPs is not from other CSPs but from already-digital global giants operating across the Internet – Facebook, Google, and Amazon in particular. To compete against these players CSPs need to adopt new technical underpinnings – such as Network Functions Virtualization (NFV) and Software Defined Networks (SDN) – to provide them the agility and flexibility to quickly deliver new and compelling services and capabilities to their customers.

The aim is to make the CSP business more flexible and agile

But simply bringing in new technologies is not enough. CSPs are realising that they must engage with partners and suppliers in a radically different way to make those technologies work for them on an ongoing basis. Open solutions and open source community participation are important elements of the technical and cultural change that are required to meet those ends. These new open, community-based approaches are leading commercial players to explore new ways of partnering with a spectrum of vendors to jointly develop and offer solutions. In this vein, Open partner ‘ecosystems’ are fast becoming a crucial catalyst and enabler of the new way forward.

Hewlett Packard Enterprise (HPE) and Intel have created a partner ecosystem along with Telco Networking Independent Software Vendors (ISVs). The objective of this ecosystem is to help accelerate NFV deployments by providing an environment and resources to enable Virtual Network Functions (VNF) testing and certification as well as foster collaboration and share learnings.

The next three big opportunities for CSPs

This first quarterly edition of CSP Evolution tracks three important technical and evolutionary strands: Our first strand examines the growing acceptance of open solutions and open source approaches for CSPs and their technology suppliers. Our second strand highlights the importance of technology supplier ecosystems as a way for CSPs to effectively leverage new NFV/SDN paradigms. And our third strand tackles NFV Management and Orchestration (MANO) and examines why MANO has become so important and where the fault-lines are surrounding its implementation.

We hope you enjoy CSP Evolution and, above all, find it useful.

The aim is to make the CSP business more flexible and agile
The next three big opportunities for CSPs

Does Open Source stack up for CSP networks?

60 seconds - Does Open Source stack up for Service Provider transformation?

It’s taken time, but open source software is now being accepted as the essential glue that will hold the NFV transformation of service provider networks together.

Can you really drive innovation through Open Source?

About four years ago it was recognised that a new way of developing telecoms networks was badly needed.

CSPs knew they had to break away from the traditional ‘waterfall’ iterative development process for network services, where you conceive, design, build, test, and implement. It was an approach which was holding them back. It was taking years to agree and then test and develop new services, when rivals – who didn’t have huge inflexible networks – could do the same thing in months.

What CSPs needed was a network which granted flexibility, agility and lower costs.

They didn’t have to look far to find inspiration and work out what direction to take.

Server Virtualization techniques developed and nurtured in the IT industry were not only well-proven and long standing, but they had helped incubate hugely successful online players like Google, Facebook and Amazon who had all exploited virtualization along with low-cost, commodity hardware platforms.

The result was hyperscale data centres and rock-bottom pricing for the sorts of services that, ideally, CSPs themselves wanted to provide or at least have a hand in.

Enter giant search engines (Google); vast global information-gathering social networks (Facebook) and hyperscale cloud facilities (Amazon – AWS) – by about 2011 the hyperscale data centres and all who sailed in them looked to be invincible. Worse, it was clear that some of their residents – like Google or Apple – were casting a beady eye at CSPs’ core network services.

For CSPs it was time to do something drastic.

What emerged was the ETSI NFV Industry Specification Group (ISG). The idea was to develop a new industry architecture for a software driven, open, virtualized network infrastructure that would enable telcos to fight it out on a level playing field by improving network agility and flexibility. And at a lower cost.

But while the need for NFV and SDN was fully accepted, less enthusiasm on the part of both vendors and (in some cases) CSPs was exhibited around the concept of open source software and open source development which, it was urged, must come as well. To many people on both sides of the fence, open source software just wasn’t the ‘right’ way of going about things. One CSP executive (no names, no pack drill) was even heard to say that open source would be adopted in his company over his dead body.

This was understandable. Telecoms infrastructure and software vendors had invested and built businesses around proprietary systems up to that point and an open source approach looked threatening.

Today things appear to have changed. The reason?

It’s now widely recognised that if you want a successful transition to NFV and a happy development environment thereafter, then open source software and its collaborative ways must play an important part in the process.

What is OPNFV’s unique role in NFV?

In fact, collaboration around open source software is accepted as perhaps the most crucial element with which the ‘virtualization’ of the network’s components can be engineered. Without mobilising a huge common effort where engineers and coders can share their knowledge and enthusiasm, a virtualized, end-to-end global network simply couldn’t be organised using the old, tried and true ‘standards’ approach.

As it’s turned out, it’s now widely recognised that open source software and an open, collaborative way of working is an important – perhaps the most important – element in the transformation.

That’s not to say that the open source community way of working suits everyone all the time.

We’ve heard about the benefits, what about the downsides?

Developers devote time and effort to open source because they’re interested in advancing things and enjoy the experience of working closely with very like-minded people. Also, according to Prodip Sen, CTO Network Functions Virtualization at HPE and chairman of OPNFV, open source is still new and there is “fear, doubt and uncertainty about it [open source] based on certain myths: that because it’s really open and ‘out there’ nobody’s really controlling it. Or that it’s unstable and that it’s buggy. The reality is that those myths are myths,” he claims.

Open source is usually a project and a community with multiple developers and so when you have many developers there are many different eyes on the code and you get intrinsic auditing, says Sen. “People are looking at it from different angles, trying out different things, discovering problems and solving them. It’s the epitome of innovation because it has diverse perspectives with a unity of vision about what’s being developed.”

Can you really drive innovation through Open Source
What is OPNFV’s unique role in NFV?
We’ve heard about benefits, what about the downsides of ecosystems?

The importance of open ecosystems for SP transformation

Digital Transformation (DX) is all the rage in the broad IT market – and it’s worth a lot of money too. According to IDC researchers, global spending on DX technologies will grow to over $2.1 billion in 2019 – that’s a rate of increase of 16% per year.

So what is ‘digital transformation’ exactly? Why might it represent more than just another large dollop of technical jargon and how does it feed into CSP Evolution story?

DX is essentially taking advantage of the blurring of the old technical demarcation lines traditionally drawn between different computer systems and functions and driven partly by the increasingly widespread adoption of open source projects and principles.

IDC defines Digital Transformation as “the continuous process by which enterprises adapt to or drive disruptive changes in their customers and markets (external ecosystem) by leveraging digital competencies to innovate new business models, products, and services that seamlessly blend digital and physical and business and customer experiences while improving operational efficiencies and organizational performance. Investments will focus on making business operations more responsive and effective by leveraging digitally-connected products, services, assets, people, and trading partners.”

Investments in operating model DX technologies help businesses redefine “how” work gets done by integrating external market connections with internal digital processes and projects

In other words, as those old barriers and demarcation lines are erased, organisations find it easier and more advantageous to connect and interlink their own internal processes with processes outside the organisation. By doing so they look to gain cost, technology, efficiency, trading and cash velocity advantages. It means that digital organisations interacting with other digital organisations can essentially rev up their joint activities since there are fewer human or other interfaces standing between them to slow the processes down.

Nowhere is the DX process moving faster or hitting harder than between equipment and software vendors and CSPs. Thanks to virtualization and open source the old technical demarcation lines are falling and closely-guarded technology and associated secrets are being replaced by vendor ecosystems where sharing and co-operation is the new normal.

In the context of NFV, ecosystems such as HPE’s OpenNFV simply make sense. It’s a structure that enables ‘best of breed’ VNFs to be tested and certified by HPE’s own labs and the whole end-to-end system of virtual functions to be integrated with the assurance that it will work at carrier grade for the CSP.

It’s working for HPE and its partners. Eight out of ten of the largest service providers use HPE’s servers and it is the global market share leader in that sector.

While the ecosystem enables a more intimate, open and ‘digital’ relationship to be forged between vendors within an ecosystem, the process also promises to change the relationship between vendors and CSPs. ABI Research, expects a shuffling of responsibilities as virtualization takes hold and operators manage the transition to virtualized telecom networks.

ABI Research anticipates that “infrastructure vendors will increase their professional service offerings to assist with this transformation, as well as managed service offerings to allow operators to focus on the other aspects of their business transformation brought on by virtualization.”

Why it’s really important for 6Wind to partner with Intel and HPE
Carrier grade very meaningful, don’t let anyone tell you otherwise
Spirent now moving to trials running in live networks, with real customers

Management and Orchestration for NFV

NFV is now well advanced, understood and appreciated, but before it can really deliver all its benefits there has to be a major effort put into finessing the so-called Management and Orchestration (MANO) – the crucial piece tasked to ‘on-board and orchestrate all those virtual functions into the agile, manageable, automated network promised by NFV.

One of the ultimate goals for virtualization and NFV from the CSP perspective is to work towards complete network automation so that what are now human-supervised processes (or even truck-rolling operations) are eliminated. To get to this ultimate goal an efficient MANO will be key.

ETSI has developed an architectural framework and core principles to define how the MANO function should manage virtualized functions in the cloud data centre. As a result there are now a range of MANO implementations and works in progress including the Open Source MANO (OSM) Group led by ETSI and the OPEN-Orchestrator (OPEN-O) project from the Linux Foundation which are both driven by open source developments.

Then there are service providers and vendors who have also developed MANO solutions including the OpenMANO project led by Telefonica and HPE’s NFV Director MANO which adheres to the ETSI MANO framework. These use open source code where they can, but that can’t always provide a workable MANO solution for a specific CSP environment by itself so, in HPE’s case, the approach is to fill in the gaps and produce a ‘vendor-driven MANO’ that can work in today’s production environment.

Two sides of a conundrum

If as an industry we’re ultimately trying to develop a compatible, interoperable environment where a virtual network function from one software developer – hatched and validated in a particular ecosystem – can work with VNFs from other ecosystems, can the industry afford to have MANO divergence which could introduce incompatibilities and integration problems over the medium term and perhaps slow progress?

On the other side of this conundrum are a host of practical considerations which makes the second, vendor-driven approach the preferred option for a growing proportion of CSPs.

It’s important to remember that management and orchestration have been proprietary functions in the past, supporting ‘native’ (black box, non-virtual) deployments. In some cases very high degrees of automation have already been achieved under the old architectural framework, so any NFV deployment now is likely to be one small element in a much bigger whole and must therefore integrate with the rest of the telco estate. That means that different MANO types will be needed to support specific CSPs through the – often lengthy – transition period.

Another factor favouring the vendor driven approach is CSP size. While large CSPs can integrate open source elements themselves, many smaller CSPs want to work with one vendor that can provide a one-stop shop for NFV transformation.

HPE’s solution

HPE is deploying its own NFV Director MANO as part of its OpenNFV architecture and ecosystem which it calculates will bring CSPs the best of both worlds: it is vendor-driven for faster deployment and unified support (one throat to choke). Its big advantage is that it will support validated, pre-integrated VNFs from the HPE ecosystem and from across the industry.

At the same time HPE’s NFV Director MANO adheres to the ETSI frameworks and will migrate with the rest of the industry towards MANO standardisation as more interfaces between different chunks of open source code are engineered.

There’s much more to understand about NFV MANO so hear what these industry experts are saying in this Intel Network Builders hosted panel discussion.

NFV is now well advanced, understood and appreciated, but before it can really deliver all its benefits there has to be a major effort put into finessing the so-called Management and Orchestration (MANO) – the crucial piece tasked to ‘on-board and orchestrate all those virtual functions into the agile, manageable, automated network promised by NFV.

One of the ultimate goals for virtualization and NFV from the CSP perspective is to work towards complete network automation so that what are now human-supervised processes (or even truck-rolling operations) are eliminated. To get to this ultimate goal an efficient MANO will be key.

ETSI has developed an architectural framework and core principles to define how the MANO function should manage virtualized functions in the cloud data centre. As a result there are now a range of MANO implementations and works in progress including the Open Source MANO (OSM) Group led by ETSI and the OPEN-Orchestrator (OPEN-O) project from the Linux Foundation which are both driven by open source developments.

Then there are service providers and vendors who have also developed MANO solutions including the OpenMANO project led by Telefonica and HPE’s NFV Director MANO which adheres to the ETSI MANO framework. These use open source code where they can, but that can’t always provide a workable MANO solution for a specific CSP environment by itself so, in HPE’s case, the approach is to fill in the gaps and produce a ‘vendor-driven MANO’ that can work in today’s production environment.

Two sides of a conundrum

If as an industry we’re ultimately trying to develop a compatible, interoperable environment where a virtual network function from one software developer – hatched and validated in a particular ecosystem – can work with VNFs from other ecosystems, can the industry afford to have MANO divergence which could introduce incompatibilities and integration problems over the medium term and perhaps slow progress?

On the other side of this conundrum are a host of practical considerations which makes the second, vendor-driven approach the preferred option for a growing proportion of CSPs.

It’s important to remember that management and orchestration have been proprietary functions in the past, supporting ‘native’ (black box, non-virtual) deployments. In some cases very high degrees of automation have already been achieved under the old architectural framework, so any NFV deployment now is likely to be one small element in a much bigger whole and must therefore integrate with the rest of the telco estate. That means that different MANO types will be needed to support specific CSPs through the – often lengthy – transition period.

Another factor favouring the vendor driven approach is CSP size. While large CSPs can integrate open source elements themselves, many smaller CSPs want to work with one vendor that can provide a one-stop shop for NFV transformation.

HPE’s solution

HPE is deploying its own NFV Director MANO as part of its OpenNFV architecture and ecosystem which it calculates will bring CSPs the best of both worlds: it is vendor-driven for faster deployment and unified support (one throat to choke). Its big advantage is that it will support validated, pre-integrated VNFs from the HPE ecosystem and from across the industry.

At the same time HPE’s NFV Director MANO adheres to the ETSI frameworks and will migrate with the rest of the industry towards MANO standardisation as more interfaces between different chunks of open source code are engineered.

There’s much more to understand about NFV MANO so hear what these industry experts are saying in this Intel Network Builders hosted panel discussion.

Will community-based MANO projects impact the proprietary status quo? 60 second video