The annual TM Forum Live conference in the sunny French Riviera is no longer the somewhat dry and specialized event it once was, but tackles issues at the very heart of carriers’ strategies for the new generation of services. Adapting OSS/BSS and network management platforms which have been developed over years, even decades, for a flexible world of on-demand applications and bandwidth is a massive challenge. At the center of this year’s discussions on this topic, predictably was virtualization, particularly NFV (the Network Functions Virtualisation specs), and the slow march to SDN (software defined networking).
Behind shiny product roadmap announcements, vendors and operators are cautious, often battle scarred from previous efforts at a ‘big bang’ transformation. Massimo Fatato, worldwide OSS domain business lead at Hewlett-Packard’s enterprise services group, told the conference: “NFV is now not a matter of if but when, but we have to be pragmatic.We cannot afford a ‘big bang’.” The change will require new skills, greater convergence of IT and network personnel, he said, and a management approach that ‘runs at the same speed’ as the ‘agility, elasticity and flexibility’ promised by virtualization.
The trend among operators is to focus initial NFV elements on packet core elements, especially those which can be transformed without significant knock-on effects on the rest of the network, and sometimes on the CPE. Some are trialling virtualized Cloud-RAN systems, but immature technologies and the LTE investment cycles suggest those remain futuristic for most cellcos. But another important area of focus is the OSS (operations support system), and Fatato argues NFV should be a catalyst and ‘a compelling event to transform OSS’.
To support such thinking, the TM Forum launched a new program called ZOOM (Zero-touch Orchestration, Operations and Management), which addresses perhaps the most important and thorny aspect of NFV, the management and orchestration of the virtual machines as they become more numerous and dynamic. Operators envisage being able to implement network functions, which currently run on dedicated appliances, as virtual machine software on off-the-shelf hardware. Many instances of a VM can be called up (and shut down) as required by user traffic, concentrating resources where they are needed and laying the foundation for on-demand services and capacity.
However, current OSS/BSS systems are ill-suited to this new world of dynamic allocation. They were developed for an era when connections were made singly and remained in place for a long time. ‘The current OSS system is broken, and it’s going to be replaced,’ Ralph Santitoro, director of strategic market development at Fujitsu Network Communications, told a conference last fall. ‘You can no longer patch it. It’s too expensive; it’s a 20-25 year old architecture; it doesn’t follow the new paradigms.’
The same sentiments were loudly voiced in Nice this week. Attendees agree that the modern OSS will have to operate in real time, its actions synchronized at all times with the network state. Another important change will be the blurring of lines between network layers, such as packet and optical, which currently work in isolation. With SDN, multiple layers may be provisioned simultaneously, as well as very rapidly on-demand.
However, there are differences on how to address the challenges, and whether existing systems ‘ which represent massive investments of dollars and effort over two decades ‘ can be adapted or need to be thrown out completely, a huge decision for a carrier which is low on time, budget and tolerance of risk. AT&T, which is in the forefront of both SDN and HetNet deployment plans, has said that management and orchestration with an SDN controller is so different that ‘this whole concept of an OSS and an EMS [element management system] all the way down, it’s actually probably archaic’. By contrast, other operators seek to adapt their current OSS platforms piece by piece; or maintain them to support existing services while using a cloud-based system to launch new ones. Some, by contrast, may even outsource legacy activities and focus their own efforts on the new OSS, either inhouse or with a hosted provider.
Some software providers are even offering the option of provisioning in an SDN-friendly manner without touching the OSS layer. For instance, Amartus provides software defined service orchestration software. This allows a carrier to describe an abstract model of a new service, and the firm’s Chameleon OS tool translates this into the language of the existing OSS.
Whatever the approach, operators will be reassured when standards mature, not just NFV itself but the surrounding interfaces. In a panel debate at the conference, Axel Clauberg of Deutsche Telekom (VP for IP aggregation, transport and fixed access in the CTO office) was in agreement with Fred Feisullin, senior network architect in the CTO office at Sprint, that ‘virtualized chaos is still chaos’, and just deploying NFV is not enough on its own. “We need to cloudify our environment, we need to fully automate and do lifecycle management,’ urged Clauberg.
Like other operators, they want standards ‘ to avoid being locked into software siloes even while breaking down hardware siloes ‘ and simplicity. The mobile carriers’ persistent but elusive goal of a flat network architecture is important in this software world too ‘ ‘we don’t want to have to orchestrate the orchestrators,’ said Feisullin, arguing for limiting the number of layers in the network. He also wants open standardized interfaces, a strong meta-model for services that can be applied across the entire network, and “tightly coupled orchestration capability” to boost efficiency and reduce running costs.
The steps towards this open world seem small, but the strength of support behind NFV is itself significant ‘ though accused of being the vendors’ way of keeping control of carrier network virtualization and ensuring it plays to their tune, NFV is nevertheless gaining a powerful momentum of its own as a full standard, because it is relatively pragmatic and meets a real operator need, and it is large enough to dilute the influence of any one player. The NFV community is now considering defining APIs between the service and orchestration layers of the network.
G’nther Ottendorfer, CTO of Telekom Austria Group, said: “We are absolutely confident that the NFV technology is a defining trend in the mobile industry’, and his words are echoed by many peers. He was speaking as VipNet, the Croatian subsidiary of Telekom Austria, became the latest in a string of carriers (and the first in the Austrian group) to demonstrate NFV functionality. It showed off LTE HD video being streamed between smartphones and the cloud over virtualized infrastructure.
Croatia has become an NFV hotbed ‘ last year, Deutsche Telekom announced it was trialling its Terastream platform in the country. This isolates the service layers and network models from the lower transport access layers, an architecture which it is initially using for OSS/BSS virtualization.
Of course, the big vendors are leaping to the challenge. At TM World, Ericsson unveiled version 2.0 of its Cloud Manager, which it says will enable operators to turn the cloud into a source of operational efficiency as well as new revenues. It focuses on enabling carriers to transition at their own pace to virtualized infrastructures and includes OSS/BSS support including orchestration, management and monitoring of services running on geographically distributed, virtualized and physical resources.
“As operators migrate from legacy solutions to the cloud, Ericsson Cloud Manager orchestrates the complex associations between virtualized network functions and underlying resources,” said Elisabetta Romano, head of OSS and service enablement at Ericsson’s support solutions business unit.
Cloud Manager is available as a standalone or managed system, showing how the big hardware makers are having to divorce their back end software from their own networks or services offerings, even as everything gets transformed into applications and VMs. Huawei and ZTE are both building up their standalone BSS/OSS offerings as well as preparing them for virtualization. In Nice. Huawei said it would provide its OSS separately from its managed services, if required, and that the software would also be available on a centralized platform in the cloud. In February, the Chinese vendor signalled that it would be more aggressive in the global OSS space when it acquired Fastwire, an Australian telecom network management firm.
It said at the same time that, while it had addressed this market mainly through its professional services unit, it now intended to build up an integrated portfolio of products to sell directly to carriers, by acquisition or inhouse development.
Meanwhile ZTE unveiled a new version of its ZSmart system and said it was now generating $537m a year in revenues from OSS/BSS, with 120 customers and a team of 3,500 dedicated R&D staff. Some of the contracts are independent of network installations, said the company, citing Orange and StarHub, both customers for ZTE’s converged policy and charging system. The new ZSmart can now be bought in modules, rather than as an all-in-one package, and it supports the GSMA’s OneAPI specifications, to encourage third party applications.