Hardiman Telecommunications: Investment Opportunities in Network Virtualisation

Moore’s Law, which states that the device density on microelectronic chips doubles every two years, is well known. Moore promulgated his law in 1965.  We now have processing capabilities that are almost seventy million times more potent than then.  Modern telecommunications switches and routers are properly considered as special purpose computers.   Implementation of functionality that was hitherto the exclusive province of hardware is transiting to software.  The resultant “Virtualisation” of network functionality promises significant cost savings in both capital and operating expenditure. In this article, Enda Hardiman, Managing Partner of consultancy Hardiman Telecommunications, considers ramifications from investment perspectives.

The basic concepts of Network Function Virtualisation have been around for some time.  They are routinely implemented in Bridged Local Area Networks and in Virtual Private Networks.  The effect is to allow users to see an accretion of nodes, PCs to take one example, as though they were locally connected.  Users will not see any difference between nodes located in the next room and nodes located on the other side of the world.

The concept of “Users” may be expanded.  Users do not need to be human.  They can be network elements, hardware or software, that require specific services.  Services may be implemented in software on switches, routers, servers, or generalised computing devices.  Requirements for expensive, special purpose, single-function hardware are reduced, indeed gradually eliminated.  The future is with powerful, relatively simple hardware driven by flexible, customisable and easily expandable software.

Terms including Software Defined Networks, Software Optimised Networks and Software Defined Radio have entered common parlance.  These are macro concepts, and are, perhaps, intuitively obvious in concept.  Each requires implementation of underlying functionality in software, according to concepts of Network Function Virtualisation.  Network Function Virtualisation is akin to concepts of Object Oriented Programming in software development.  Instead of defining software modules according to procedure of operation, they rely on defined processing characteristics : Any given set of inputs will produce robustly predictable output in precisely defined format.  As will be obvious, rigorous specification and standardisation of Virtualised Functions is essential. To reverse the acronym, Virtualised Network Functions define a set of well-specified implementations which, when linked together, define and deliver network services.

In these days of fast fibre interconnection of global scope, there is no overriding technical reason why Virtualised Network Functions need to be implemented in close adjacency.  They can be distributed, in fact dispersed globally if required. This however carries two caveats.  The first is with respect to latency.  Some network operations must complete in timescales measured in milliseconds.  Latency – delay – over fibre may be unacceptably long.  The second is with respect to traffic volume generated.  In any given instance, the question as to whether costs of remote processing, using transmission over fibre, are higher or lower than costs of local processing has a deterministic answer.  The deterministic answer allows the respective CapEx and OpEx requirements of competing implementations to be assessed.  Thus is defined the fundamental logic of deployment of specific virtualised architectures.

Standardisation, alluded to above, is an iterative process.  It has been an invariable feature of all generations of telecommunications technology that while revolutionary concepts may be agreed upon industry-wide, there follows a period of diverse implementation by equipment manufacturers and software suppliers.  Industry-wide adoption of very precisely specified functionality and interfacing requirements takes some time.  Ultimately, it occurs under the aegis of standards bodies.  In the interim, however, commercial demand dictates that individual manufacturers and software suppliers proceed with efficient, if, arguably idiosyncratic implementations that bring about the desired benefits.  It is generally possible to achieve interoperability of equipment and software from different sources by implementation of customised “Bridging” software functionality; the facilities thus constructed go by the name of Mediation Devices.  Telecommunications operators routinely specify and buy them.  Mediation Devices can be simple or complex, and, again, simplicity or complexity can have major ramifications on CapEx and OpEx.

Equipment manufacturers do not produce all of their own software.  Software is frequently sourced from independent software suppliers.  Selling through equipment manufacturers is advantageous to software suppliers.  Systems integration, that is, integration of the offered software products with existing facilities, including other Virtualised Network Functions, is undertaken with the assistance of manufacturer architects and programmers familiar with specific equipment requirements and with network-wide functionality.  Equipment manufacturers are deeply experienced in this area.  They have to be, in order to achieve the “Five Nines” – that is, 99.999% reliability - demanded by telecommunications operators.  Reputable manufacturers have exceptionally rigorous concepts of definition of the precise dimensions and specifications of each System Build – that is, each hardware and software combination.  Testing, including regression testing, is facilitated Thus, telecommunications operators can be confident that a software product sourced through a reputable equipment manufacturer is robust and stable.  From the software supplier’s perspective, development cycles and integration testing may be somewhat protracted, but in the overwhelming majority of cases this is more than compensated for by reduced warranty and post-sales support OpEx.  Those OpEx elements, being of their nature difficult to forecast with precision, can have major impact on the commercial performance, indeed viability, of software suppliers.

Selling via manufacturers is not always possible.  Larger telecommunications operators usually have R&D facilities that are directed at implementation of advanced functionality that is required in advance of availability in standard equipment manufacturer offerings.  Software suppliers thus also sell directly to telecommunications operators.  One advantage of this approach is that it allows early visibility on trends that will acquire commercial momentum, and thus facilitates characterisation of product roadmaps accordingly.  That can be significantly advantageous in maintaining technical differentiation and competitive positioning.  It is however fraught with one major risk, to which it is all to easy to succumb.  That is the risk of deployment, by the software supplier, of an excessive number of multiple, slightly or significantly different “Tailored” implementations, each developed against the specific requirements of an individual operator.  Post-sales support, maintenance and warranty OpEx rapidly spin out of control.  Software suppliers need to be very clear, and very ruthless, concerning that which defines core functionality and that which defines specially tailored implementations. Mixing the two entails major risk of instability and, as noted, uncontrolled OpEx.  Revenues derived from “Tailoring” may by all means be allocated to development against standard product roadmaps, but seeking to define the roadmaps against customised implementations is rarely, if ever, advisable.

Other aspects of the business models of software suppliers merit consideration by investors.  Technical due diligence of course must be undertaken, and should feature code walk-through, stability of application programming interfaces and other technical arcana.  However, it is actually quite rare to encounter early stage software operations that exhibit catastrophic technical lacunae.  Some stability engineering may be warranted, but more often than not core concepts will be found to be relatively robust.  However, early, and even mid-stage software suppliers considering market demand forecasts and competitive positioning may confuse technical innovation with marketability.  Innovation must be matched with economics as considered by telecommunications operators.  Does the product do one or all of reduce CapEx, reduce OpEx or sustain increased revenue streams ?  To what extent can these be quantified ?  Technical differentiation alone will not suffice.  The matter of intended audience is also important.  What feature of the product need to be emphasised to technical people, to operations people and to commercial people amongst potential customers ?

Cash flow forecasts of software suppliers always merit close investigation.  It is very common during the early stages of software company development for promoters to be exaggeratedly optimistic with regard to sales in general, and to the length of sales cycles in particular.  Ramifications on cash flow, most particularly on working capital requirements need rigorous scrutiny.  Points made above regarding realistic casting of post-sales warranty support OpEx merit reiteration here.  There is often a strong temptation, which it is of paramount importance to resist,  to release inadequately tested software in order to meet ill-considered commitments that were originally made to secure a sale.  Customers may in time forgive delayed delivery, but they will remember unstable products for a long time.

The market for supply of Virtualised Network Function products is quite fragmented at this time.  There is without doubt an abundance of investment opportunities.  Some of these will be very lucrative for investors.  There will however be a rationalisation of suppliers, in the form of insolvencies, mergers, acquisitions, and the emergence of industry leaders that will, so to speak pull away from the also-rans.  Those likely to feature in the ranks of the leaders are, even now, those which exhibit realistic approaches to channels to market, well-defined processes of product and product profile control, accurate understanding of metrics of the competitiveness of their products, thorough understanding of OpEx calls forward, and, not least, realistic casting of working capital requirements.

 

Enda Hardiman is Managing Partner of Hardiman Telecommunications Ltd. (http://www.telecoms.net).  Hardiman Telecommunications is a boutique consultancy with an extensive practice in strategy, M&A and performance enhancement in Europe, the Middle East, Africa, Asia and the Pacific Rim. He may be contacted at hardiman@telecoms.net.