Moreover, the operator still has a long way to go on its transition to all-IP networks, which Clauberg sees as a prerequisite for the rollout of software and virtualization technologies. Deutsche Telekom aims to complete the all-IP transformation by 2018, but only half of its fixed lines in Europe (excluding Germany) were IP-based in March, up from about 45% in September last year. PSTN networks have been entirely switched off in the smaller markets of Croatia, Macedonia, Montenegro and Slovakia, but they still feature prominently in the much bigger territories of Germany, Greece and Romania.
Table 1: All-IP Targets & Progress
|Country/region||Total fixed lines in March ('000)||IP fixed lines in March ('000)||Percentage all-IP (March)||Status/target for completion|
|Source: Deutsche Telekom|
Deutsche Telekom has previously indicated that it costs between €30 ($30) and €60 ($68) per customer to make the all-IP switch, putting the overall investment needed in Europe and Germany at between €863 million ($979 million) and €1.73 billion ($1.96 million). With 59% of that investment yet to be made, the spending requirement works out to be somewhere between 12% and 23% of Deutsche Telekom's capital expenditure outside the US market in 2015, minus payments for spectrum licenses. Given the eye-watering investment needs at its access networks division, this will be tough to bear.
Ultimately, though, Deutsche Telekom reckons it can save about €10 ($11.4) per customer annually from all-IP transformation. That would put savings at about 2% of EBITDA in Europe and 2.4% of the figure in Germany last year. But the rollout of SDN and NFV should lead to even bigger savings in future, says Hischke. According to a white paper published last year by management consultancy Arthur D. Little and equipment vendor Alcatel-Lucent (NYSE: ALU) (now subsumed into Nokia Corp. (NYSE: NOK)), a European operator could reduce operating expenditure by a tenth through investment in SDN and NFV. (See Will Investing in SDN & NFV Be Worth It?.)
Table 2: Deutsche Telekom Versus European Rivals on All-IP
|Operator||All-IP deadlines||Progress to date||Publicly stated goals/benefits||Notes|
|BT||2025||All-IP core network already built||Simplifying estate; replacing legacy networks; saving costs||Operator claims 21CN project achieved all-IP-to-exchanges target and that current initiative will take all-IP to premises|
|Deutsche Telekom||2018||Croatia, Macedonia, Montenegro and Slovakia are all-IP||Service agility; cost savings (€1.2 billion ($1.3 billion) by 2020 from use of cloud-based technology)||All-IP move is designed to support a major reduction in the number of service platforms Deutsche Telekom maintains across Europe, from about 650 before the process began to just 50 at its conclusion|
|KPN||N/A||>60% of connections||Optimization for very high data volumes; reduced complexity; faster time to market||To support our best-in-class access services, we continue to move towards an all-IP network, said KPN in 2014 annual report. "Moving towards a single IP core is needed to ensure best-in-class integrated services"|
|Orange||N/A||All broadband services in Mauritius are all-IP||Improved user experience; better scalability, security and energy efficiency; enabling development of new services||All-IP pilot under way in Mauritius, which was chosen, says Orange, because of its "size, the representativeness of the different markets, such as residential, enterprise and mobile, and the presence of all services and technologies"|
|Swisscom||2017||>25% of connections; >33% of customers||Enabling cost-effective use of new services; offering services irrespective of type of access technology||Has migrated data transport network to IP, commissioned IP-based telephony and multimedia platform and been offering IP-based services since 2009|
|Telecom Italia||N/A||N/A||N/A||Migration from PSTN to all-IP under way|
|Telefónica||N/A||N/A||N/A||All-IP forms a part of network transformation Telefónica is carrying out|
|Telekom Austria||N/A||All fixed voice customers in Austria on all-IP||To overcome the problem of technology obsolescence related to the use of legacy networks||Has revealed that another subsidiary will shortly make an announcement on all-IP conversion|
|Source: Goldman Sachs, operators, Light Reading|
Deutsche Telekom has not provided any detailed guidance here. That is possibly because the immaturity of these technologies makes it hard to predict what impact they will have. Moreover, a recent survey of telcos by Heavy Reading suggests that expectations about cost savings are in constant flux. "Not everything is ready and so we have to use interim solutions and have a clear migration path for each and every platform," says Hischke. Using legacy hardware, the operator has already developed a pan-net messaging platform that supports SMS and MMS traffic. (See Opex Gloom Grows in New NFV Survey.)
Apparently frustrated by slow progress on the design of non-proprietary technologies, and eager to escape the clutches of some vendors, Deutsche Telekom has taken a leading role in Facebook's recently launched Telecom Infra Project (TIP), whose mission is to speed up the development of open-source technologies. Clauberg, who now occupies a board position at TIP, has also complained about the poor scalability of first-generation NFV. "Most of us start with virtual machine-based virtualization, which is very heavy… and doesn't scale endlessly," he told the audience at BCE. "Containers seem to be a better way to go for some applications." (See Facebook: TIP Will Open Telecom Hardware.)
In this kind of set-up, or one based on the closely related concept of "microservices," a network function would be decomposed into small individual components that a service provider could reuse in different ways to create customized, scalable applications. Caroline Chappell, practice leader of cloud and NFV at Heavy Reading, has described microservices as "real cloudification" and said they will deliver more extreme business benefits to operators than first-generation NFV, including greater automation and the ability to introduce new services much faster.
That could be critical for Deutsche Telekom. During trials of pan-net offerings carried out in Croatia, Hungary and Slovakia last year, the operator was able to set up a VPN of "medium complexity" in just 15 minutes, according to Claudia Nemat, Deutsche Telekom's Europe and technology boss. Previously, that process could take anywhere between two and four weeks. Clauberg has recently emphasized the importance of accelerating other service development if Deutsche Telekom is to avoid ending up as a dumb pipe. "Internet companies are using software-based production and have short market lead times and can take products off the market very quickly," he said during the MPLS/SDN/NFV World Congress in Paris in March. "We need to play the game with the same tools." (See DT: We Need SDN, NFV to Battle Web Giants.)
Yet Clauberg has acknowledged that containers may not be the answer for everything. According to Peter Willis, the chief researcher of data networks for the UK's BT Group plc (NYSE: BT; London: BTA) (now 8% owned by Deutsche Telekom), microservices could actually increase complexity and lead to performance problems by exposing more interfaces externally. What's more, Heavy Reading's Chappell reckons the investments most equipment vendors have made in existing products give them little incentive to support microservices deployments. Perhaps with such thoughts in mind, Clauberg complained that operators are living in a "vendor prison" during his talk at BCE. (See The Real NFV Revolution Is 5 Years Away and DT: Telcos Must Escape Vendor Prison.)
Next page: People problems