Traffic Management
Content
Traffic management: six guiding issues
Six issues guiding discussions on traffic management
Over the last year I have been blessed to be able to avoid dealing with "5G."
(I suspect that I will run out of luck in the coming year, but maybe someone
will be able to tell me what 5G is by then, and that will make it more
relevant.) Instead, I have been working on different topics that, in retrospect,
seem to have a common focus: how to make the best use of the resources
that are available already. Among them, traffic management has a prominent
role.
It is easy to see why talking about new technologies like 5G, or major LTE or
LTE Advanced rollouts is a more grandiose and sexy conversation topic.
Managing traffic, optimizing network utilization, and improving performance
and quality of experience at the margin sound like--and is--hard work.
But it also opens a fascinating perspective on how to run and extract the
most from mobile networks. The challenge of how operators can pack more
traffic and services on the infrastructure it already has and improve
subscribers' experience is not a new one--but there has been great
innovation in the tools, their availability and their effectiveness over the last
few years. Our understanding of how traffic dynamics have changed and
continue to change, and how to track them in live networks makes the case
for active traffic management much more compelling from a cost-saving and
revenue-generating perspective.
Even more important, however, is the fact that many mobile operators are
now fully onboard and they recognize that this is a priority. A couple of years
ago, skepticism and doom-and-gloom was a far more common reaction to
any solution to relieve congestion or improve network efficiency.
If you think of traffic management as the task of moving IP packets through a
mobile network as fast as you can, it sounds like a rather pedestrian effort.
True, scalability and speed are anything but trivial, but they can lead to a
one-dimensional race among vendors first, and operators next, in which the
subscriber experience may end up in the background.
But mobile traffic is much more than IP packets--and capacity much more
than end-user downlink speeds. The forecasts that show the massive
increase in mobile traffic are only the starting point in a much more intricate
and consequential story that average traffic estimates cannot capture.
There are multiple dimensions in traffic: subscriber, devices, application,
service, traffic type (voice, video, texting, etc.), location, Radio Access
Network conditions and policy, to name a few. They not accidental features;
they define the requirements, and the can be used by the operator to
optimize the use of the existing capacity they have.
Because network resources are necessarily limited by factors like physics,
technology, spectrum availability, need for operators to be profitable, and
subscribers' willingness to pay, it is crucial to use network resources as
efficiently as possible. Increasing capacity through network expansion and
densification is necessary and useful, but not sufficient.
Of course, having to take into account all these additional dimensions when
managing traffic increases the complexity of the task over simply routing IP
packets. So there are tradeoffs that operators have to assess to decide which
solutions bring a positive contribution in terms of performance, QoE or total
cost of ownership. But far from being a burden, resource constraints and
traffic differentiation should become--and increasingly are--a springboard to
differentiation and innovation.
What does all this translate into new directions for mobile operators as they
take the lead in taking control over and benefitting from traffic complexity?
Here are six complementary trends that I encountered in my work and that
seem to shape the discussion now:
Real-time traffic management to adapt to highly dynamic traffic
patterns
Context-dependent services to use network resources to maximize QoE
Application-layer traffic management to exploit traffic variability to
improve QoE and efficiency
Virtualization to provide the flexibility needed to develop and test new
services and traffic management tools
Analytics to understand current use of network resources and scope for
optimization
Continuous network testing, monitoring and optimization to stay ahead
of changes in usage patterns.
These trends are all tied to--and reinforce--each other in providing the tools
to finally learn how to best leverage traffic dynamics to increase network
utilization and QoE. All the tools supporting these efforts have flexibility at
their core, which means that operators have the freedom to explore different
paths. We do not know which paths are the best to follow yet, and there is
much to learn. We may not have the answers yet, but we have the tools to
ask the right questions.
Six issues guiding discussions on traffic management
Over the last year I have been blessed to be able to avoid dealing with "5G."
(I suspect that I will run out of luck in the coming year, but maybe someone
will be able to tell me what 5G is by then, and that will make it more
relevant.) Instead, I have been working on different topics that, in retrospect,
seem to have a common focus: how to make the best use of the resources
that are available already. Among them, traffic management has a prominent
role.
It is easy to see why talking about new technologies like 5G, or major LTE or
LTE Advanced rollouts is a more grandiose and sexy conversation topic.
Managing traffic, optimizing network utilization, and improving performance
and quality of experience at the margin sound like--and is--hard work.
But it also opens a fascinating perspective on how to run and extract the
most from mobile networks. The challenge of how operators can pack more
traffic and services on the infrastructure it already has and improve
subscribers' experience is not a new one--but there has been great
innovation in the tools, their availability and their effectiveness over the last
few years. Our understanding of how traffic dynamics have changed and
continue to change, and how to track them in live networks makes the case
for active traffic management much more compelling from a cost-saving and
revenue-generating perspective.
Even more important, however, is the fact that many mobile operators are
now fully onboard and they recognize that this is a priority. A couple of years
ago, skepticism and doom-and-gloom was a far more common reaction to
any solution to relieve congestion or improve network efficiency.
If you think of traffic management as the task of moving IP packets through a
mobile network as fast as you can, it sounds like a rather pedestrian effort.
True, scalability and speed are anything but trivial, but they can lead to a
one-dimensional race among vendors first, and operators next, in which the
subscriber experience may end up in the background.
But mobile traffic is much more than IP packets--and capacity much more
than end-user downlink speeds. The forecasts that show the massive
increase in mobile traffic are only the starting point in a much more intricate
and consequential story that average traffic estimates cannot capture.
There are multiple dimensions in traffic: subscriber, devices, application,
service, traffic type (voice, video, texting, etc.), location, Radio Access
Network conditions and policy, to name a few. They not accidental features;
they define the requirements, and the can be used by the operator to
optimize the use of the existing capacity they have.
Because network resources are necessarily limited by factors like physics,
technology, spectrum availability, need for operators to be profitable, and
subscribers' willingness to pay, it is crucial to use network resources as
efficiently as possible. Increasing capacity through network expansion and
densification is necessary and useful, but not sufficient.
Of course, having to take into account all these additional dimensions when
managing traffic increases the complexity of the task over simply routing IP
packets. So there are tradeoffs that operators have to assess to decide which
solutions bring a positive contribution in terms of performance, QoE or total
cost of ownership. But far from being a burden, resource constraints and
traffic differentiation should become--and increasingly are--a springboard to
differentiation and innovation.
What does all this translate into new directions for mobile operators as they
take the lead in taking control over and benefitting from traffic complexity?
Here are six complementary trends that I encountered in my work and that
seem to shape the discussion now:
Real-time traffic management to adapt to highly dynamic traffic
patterns
Context-dependent services to use network resources to maximize QoE
Application-layer traffic management to exploit traffic variability to
improve QoE and efficiency
Virtualization to provide the flexibility needed to develop and test new
services and traffic management tools
Analytics to understand current use of network resources and scope for
optimization
Continuous network testing, monitoring and optimization to stay ahead
of changes in usage patterns.
These trends are all tied to--and reinforce--each other in providing the tools
to finally learn how to best leverage traffic dynamics to increase network
utilization and QoE. All the tools supporting these efforts have flexibility at
their core, which means that operators have the freedom to explore different
paths. We do not know which paths are the best to follow yet, and there is
much to learn. We may not have the answers yet, but we have the tools to
ask the right questions.
