In conjunction with Chris Lewis (Lewis Insight), I've written a fairly provocative white paper for Amdocs, looking at whether the hype about "app-aware networks" and policy management is realistic. It's one of a series aimed at telco CTOs & CMOs, done by different analysts - see here for a link to all of them. A quick intro video is below.
Much current discussion in the network industry
concerns topics such as virtualisation, new approaches to monetising data
traffic, or the shifting balance-of-power between application providers and
infrastructure owners. Despite the furore around Net Neutrality, there
continues to be a desire to treat Internet traffic differentially, as well as
creating other new non-Internet capabilities in the network, perhaps specific
to IoT connectivity or on-net content services.
Although innovations like SDN (software-defined networks)
and NFV (network function virtualisation) are being associated with
“network slicing” and “application-awareness”, those general concepts have
already existed in similar form for some years. Both fixed and mobile networks
have sought to offer so-called “value-added” functions, typically through boxes
in the data path that discriminate between different flows, alter certain
traffic’s paths or content, or add in additional data in some fashion.
Yet there is a growing tension here between stakeholders –
and also a disconnect in terms of mutual understanding and terminology. Often,
the network attempts to act directly on data and applications transiting it –
whether that is in terms of QoS management, assurance and priority (where
legally permitted), or more mundane actions such as content-optimisation and
filtering. Yet at the same time, the nature of applications is itself evolving
fast.
Apps are increasingly-complex amalgams of multiple web and native
components; they are able to adapt their behaviour to the network conditions,
avoid scrutiny by encryption or obfuscation, and change their nature overnight
by merging – or splitting into separate services. The way that an application
is perceived by the network may be very different to how it is perceived by a
user. Network architects and operations staff do not always look at the
situation “through a developer’s eyes”.
These differences do not suggest the hoped-for utopia is
realistic, with telecom operators obtaining significant incremental revenues
from content or application providers, arising from extra value from
network-awareness. Neither does it signify that traffic-management will be able
to moderate congestion or balance network loads as much as desired. If
anything, it perhaps suggests the opposite, as developers look to circumvent
network restrictions or unwanted intervention – for example by ever-greater use
of encryption, pushing traffic over 3rd-party WiFi, or by alerting
users and regulators to unwanted policy-management actions. Against that
backdrop, some of the rhetoric from the telecom industry about so-called “OTT”
providers is unhelpful and antagonistic.
Disruptive Analysis is even aware of an incident in which a network operator accidentally degraded one of its own VoIP applications - the app development team and network policy-management team didn't realise what each other were doing.
A new white paper published by Amdocs, and written by
Disruptive Analysis and Lewis Insight hopes to shed more light on this
dichotomy, and suggests ways to bring together the network operators and their
peers in the application space. There certainly are ways to link networks and applications, but they will need much
tighter collaboration and a focus on “mutual consent” as a primary goal, rather
than monetisation.
In particular, it is important that CTOs and network
architects gain a deeper understanding of how applications are evolving, and
the tools at developers’ disposal to help create higher-quality experiences.
While the network is important, so are devices – and so also are any business
processes involved in optimising performance. The average software developer is
dis-incentivised to work more closely with network operators if they face
sizeable practical hurdles or costs, for little extra provable benefit.
One important priority should be for networks to focus on
providing some sort of information and status feed to applications. If
developers can see that there is current or possible congestion or other
problems, they can tune applications so that they mitigate the issues, rather
than exacerbate them. It may be easier for the network to offer a “weather
forecast” to application and content providers, rather than trying to control
the rain and sun.
The App-Aware Networks & Network-Savvy Apps white paper
is one of a series published by Amdocs + Lewis Insight on network-related thought
leadership, written by a range of independent analysts & specialists. The paper can be downloaded here, along with video clips and summaries of the other papers in the series.
No comments:
Post a Comment