Virtualization Technology News and Information
Metaswitch 2020 Predictions: Telcos "get" Cloud in 2020

VMblog Predictions 2020 

Industry executives and experts share their predictions for 2020.  Read them in this 12th annual series exclusive.

By Paul Brittain, VP Product Strategy, Metaswitch

Telcos "get" Cloud in 2020

Technology predictions are like chili in cooking - you need to add only just the right amount of controversial opinion if the dish if to be interesting, yet remain palatable.  Let's see whether I can achieve that, with three little pre-Christmas crackers from my 2020 crystal ball:

1)      GitOps will arrive in Telcoland

2)      Telcos adopt Public Cloud

3)      ...but containerization won't arrive for telco until 2021

Each of those deserve a slightly more considered look.

GitOps will arrive in Telcoland

Telcos are notoriously conservative when it comes to adopting new technologies, and none more so than their ops teams.  That's with good reason, though, as we all rely on the networks remaining stable even in extreme overload.  Put another way, I am sure you all just assume you can always call for help when your house is on fire.

Up to now Telco automation has been dominated by variants of scripting. Think Ansible, HEAT, etc.  Great technologies, but they are all essentially a means of encoding appliance-based maintenance operation protocols (MOPs). Useful, but not game-changing because of an inherent fragility when the state of the network/router/SBC/VNF/etc does not match what it is supposed to be.  Yes, I know that you can improve robustness by forcing a rebuild from a known state, but what happens to those emergency calls in the meantime?

GitOps is genuinely different.  It is based on declarative models of the network or service state, with each element able to iterate toward the desired state, which is mastered in Git and automatically pushed out across the network when it changes. No more MOPs. No secure shelling (SSH'ing) into individual appliances - just twist the desired dial and let the pipelines do the work.  Better yet, any components that are temporarily down or unreachable will be automatically recovered. The 60%+ of service events caused by accidental config errors are eliminated, and even if the worst happens a simple button press will revert the whole network to the previous good state.

"What about ETSI MANO?" I hear some cry.  Actually, that fits well with GitOps, for example using pipelines to manage software levels and config of the VNFs making up a service in combination with MANO-derived service-level modelling and orchestration of NFVI resources and surrounding physical network functions.  Whether the VNFM logical functions continue to be provided within the MANO domain or are replaced by Kubernetes Operators is a debate for another day...but at Metaswitch we are seeing huge interest from Telco operators in declarative GitOps-based config management as the evolution of their earlier focus on lifecycle management.

Telcos adopt Public Cloud

Possibly the spiciest dish on today's menu is this second item, but I believe it will happen.  Telcos see benefits in outsourcing the additional operational complexity that the cloud brings.  The Web-scale cloud providers want telco customers, not least as they look forward to future collaboration on 5G and edge compute applications.  But these dance partners have not come together to date.

2020 will change that.  Pricing has dropped to the point where telcos can no longer simply dismiss public cloud.  Crucially, the Web-scale players have built ever more performant options in their cloud and have invested heavily in securing the environments they offer to fussy government or telco customers.  Net is that although some still say custom HW/SW infrastructure is necessary to host high-performance VNFs (most often their own pet-like products, of course), at Metaswitch we believe generic cloud works fine.

The gentlest path for introducing telco to public cloud will be to move some of their vast data lakes for diagnostics and analytics to cheap, reliable storage in the cloud, and this will probably be their first move. The right opportunity will be sufficiently self-contained and well-understood to allow both parties to perfect the tricky footwork needed to satisfy telco lawyers and regulators as well as the engineers.  More comprehensive use of public cloud for core network functions, whether day-to-day, as a DR option, or using telco real-estate in massively distributed edge compute JVs, will follow once these early steps have been perfected.  

If you are tempted to cry foul because I predict "only" cloud storage use in 2020 or wish to point to vCPE deployments by telcos as evidence that it has already happened, please note that I am referring here to core network usage of public cloud.  The regulatory implications of that is what scares telco design teams and is a much bigger step than hosting of a few niche edge services.

Containerization for Telcos comes in 2021

Many telcos are already making use of containers on their private clouds, typically via VNFs which the vendors may have developed as containerized microservices - as we do at Metaswitch - but which are packaged up, consumed and managed by the telcos today as pre-built VMs.  Similarly, recent versions of OpenStack control plane are extensively containerized, even if the telco is not explicitly managing the underlying containers. 

Large-scale production use of containers as first-class citizens in telco clouds is technically feasible today, but I don't think it will become mainstream in 2020 for two reasons.

First, the right combination of technology has only just become available.  There has been excellent progress during 2019 with mainstream distros now offering Kubernetes support for multiple network interfaces, CPU pinning and SR-IOV, Kubernetes Operators as part of GitOps (see 1 above) to ease management of complex network functions and Kubevirt (currently in tech preview from Red Hat) for hosting VM-based payloads on a container-based platform.   However, with so many new concepts and capabilities, it will take time for telco teams - especially those concerned with business continuity and security - to be comfortable putting critical infrastructure on these platforms.

Secondly, and more tellingly, the real operational benefits of containers for the telcos come after they move from an appliance-based world view to a holistic cloud-centric service-based model. Like all such mindset shifts, it takes elapsed time because it requires change in people's work habits, not just availability of the enabling technology.  Native container platforms and VNFs will appear in telco labs in 2020, but real production use will take a little longer for this reason.

Netting all the above down to one paragraph:  Telcos want to adopt containerized cloud with the latest and most efficient operational tooling, and they want to make use of the scarce expertise that Web-scale cloud providers already have in running such networks. Large-scale use of 5G slicing will be the driving force - especially for more novel use cases than just faster mobile broadband.  2020 will be the preparation year for these seismic shifts, as telco ops teams become familiar with GitOps, plus the legal and security teams get comfortable with use of public cloud.

Whether you view that as too little or too much spice will likely betray whether you are a cloud technologist or a NOC operator... I look forward to your comments and feedback.

Right or wrong, vindaloo or korma, I hope you enjoy the new challenges that 2020 brings your way!


About the Author

Paul Brittain, VP Product Strategy, Metaswitch

Paul Brittain 

Paul Brittain has more than 30 years of experience in data and telecommunications protocols, products and network design from SDLC links to multi-national cloud IMS architectures. As Vice President Product Strategy for Metaswitch, he co-ordinates future direction for the Metaswitch portfolio, including IMS, VoLTE and NFV solutions for mobile and fixed operators with a particular focus on Cloud Native architectures and operational automation.  

Paul joined Metaswitch in 1986 after studying mathematics at Cambridge University. He is the author of, or major contributor to, several IETF RFCs, multiple white papers and patents covering everything from protocols deep in the network to innovative user experiences for multi-modal communications and is a frequent speaker at conferences on NFV and orchestration. When not at work, he can most often be found sailing around the Scottish islands.

Published Friday, December 13, 2019 7:36 AM by David Marshall
There are no comments for this post.
To post a comment, you must be a registered user. Registration is free and easy! Sign up now!
<December 2019>