Exponential increases in network complexity coinciding with an ongoing reduction in skills means that networks need to be as efficient and low cost as possible whilst maintaining high performance. However, virtualisation, which has gained momentum in 2019, hasn’t lived up to expectations. It is not delivering on the promise of transformed costs and has the negative effect of introducing unnecessary operational risks. Large telco providers face a stark choice; to radically change, or risk irreversibly falling behind market competitiveness.
As CTOs are under increasing board pressure to deliver a viable and accelerated path to the cloud, to maintain a competitive edge, and to transform operations methodologies to take advantage of Dev Ops efficiencies, they are evaluating bolder and increasingly more radical strategies to move to cloud native architectures and to target extreme automation objectives.
The race to cloud native takes on even more urgency with the emergence of radical ‘enterprise disruptors’ such as Rakuten, an e-commerce player in Japan, who took advantage of their AI/enterprise expertise/philosophy to change the whole game and launch a greenfield cloud native network.
The traditional incumbent suppliers, at both the network and SI levels, sense both a significant risk to their core revenues, and an opportunity to extend from their core outsourced landscapes and to gain new territory. Smaller agile players are circling to replace physical assets with greenfield cloud native apps, and the Public Cloud players stand to profit from the transformations in all scenarios.
So why are virtualized networks not achieving the expected results? Telecoms is one of the last markets to be virtualised, due to legacy complexities and NEP strategies which protect their traditional revenue streams by maintaining closed technology stacks and minimal compliance on standards. While virtualisation represents the first logical step taken by many operators, the business case and its relatively modest returns, (10-15% on infrastructure costs and 10-25% achievable automation), are increasingly being questioned.
Ben
McCafferty,
SVP of Strategic
Alliances,
Accanto
The main reason for the low returns is the inherent inefficiencies that virtualised networks introduce. These include:
Infrastructure - While the operator is able to purchase more open/bulk hardware and attract price-performance benefits, the simple act of deploying multiple software solutions designed to operate on physical "tin" configurations in a virtualised environment, introduces inefficient infrastructure utilisation.
Software - Software is migrating to a virtualised network; the legacy software is moved in a "lift and shift" mode onto shared infrastructure, together with its OS and third-party software. This still means that OS licensing costs remain and introduce risk because software is often not proven, or effective in that environment, nor is it reducing the licensed total cost of ownership.
Hypervisor - A mediation layer called a hypervisor which sits between the software and hardware which is required to ensure the software operates in a virtual environment. This introduces additional latency and cost because the software is being deployed in a not-for-purpose environment which offsets both performance and cost gains.
#1: ACHiEVING transformational, intelligent networks
The first prediction. In 2020, due to the competitive environment, service providers will be focussed on accelerated application migration to Hybrid Network Cloud Native architectures, which are driven by 5G/IoT/Edge rollouts, and also the requirement to consolidate traditional datacentres which represent enormously complex initiatives, given the legacy environment and requirement to maintain operational services at all times.
These Cloud Native architectures are less telco-specific and deliver transformational change through optimised hardware/software operation, combined with advanced automation tools. The main challenge for service providers will be to ensure the orchestration/automation tools have been purpose-built to manage cloud-native environments and Kubernetes standards as well as supporting the traditional physical and virtual architectures, thus enabling a rapid and phased upgrade to a cloud-native enabled platform.
#2: THE MOve TOWARDS "CLoud-native ready"
The second prediction is that in order for software providers to survive they will all have to provide 'cloud-native ready' software versions which are re-architected to take full advantage of open standards.
#3: maturation of NFV and VIrtualisation
The third prediction is that in 2020 new enterprise network suppliers will finally be able to provide a solution that is incredibly competitive with massively improved price-performance, and the hurdles that we've experienced with NFV and virtualisation will be a thing of the past. If all of the considerations above are taken into account, the benefits are projected at 30-50% for infrastructure via increased density, and automation levels can finally reach 90%.
