coverstory_cover story 26/11/2014 18:57 Page 4
maintained up-to-date, disaster recovery etc.
Provided a cloud based solution meets your
data protection requirements then a cloudbased solution could be advantageous. For
Tier 1 operators, a self hosted and managed
solution is still likely to offer the best
economies of scale and allow
development of the system to meet
specific requirements.
JDSU: To the cloud of course! To be
fair, T&M use cases have shifted to
the cloud of their own accord. JDSU
has implemented cloud based
workforce efficiency solutions, for
instance. But to address the specifics
of CDN, which is basically what
video moving to the cloud is, T&M
inherits several new requirements.
There’s a need to confirm that the
video files that sit on the cloud are
valid. A need to confirm the ABR
transcode process is good, a need to
validate the signalling and
authentication tied to each video service. And
a need to correlate network impairments to
service QoE issues so we can properly
segment the network.
Mariner: In the absence of centralised
delivery, content and services are served
from multiple points, and delivered over
multiple possible network infrastructures. So
in addition to the more traditional
monitoring capabilities associated with
services and network infrastructure, useful
insight can best be provided by triangulating
content consumption backwards from the
consumer to specific delivery paths and
content sources,. The cloud increases the
be in the cloud. That is why we introduced
our own cloud QC solution, QCloud
specifically for cloud-based workflows.
Torque: Moving to the cloud presents some
special problems. But it depends on how the
delivery is moving to the cloud. Tremendous
savings can be made by moving playout
systems and management to cloud based
services. Instead of paying up front capital
costs (with fixed depreciation over several
years) operators can enjoy ever declining
monthly costs. So, if you move the traditional
encoding and playout systems into the cloud,
the exact same T&M requirements would
move with them.
VeEX: There is not really a shift with it,
rather an additional piece to monitor. The
cloud is a big word that everyone likes to use
because you are cool when you are in the
cloud, however, if the cloud does not connect
to your home, then the cloud is pretty
service. This requires powerful solutions for
aggregation, correlation, visualisation and
analytics.
Bridge: Yes. The key thing is to use the right
amount of monitoring for the system to be
reliable, and to determine what is the right
amount? We must acknowledge that
monitoring all parts of the network
may not be feasible or necessary in
most cases. If you monitor a CPE
device such as a Smart TV and
there’s an outage, the CPE metrics
must be watched with other
monitoring equipment upstream so
that data correlations can be made
to help diagnose the cause.
Farncombe: The price of the CPE
should be no barrier to monitoring
QoE. There is more of a challenge
with unmanaged devices, as
monitoring will be dependent on
what QoE interface the device
makes available to a solution.
Standardisation has helped this process but
where devices offer only limited QoE data
this can necessarily limit what can be
understood about the end user experience. If
the operator is managing the network this, to
an extent, could be mitigated by extensive
monitoring at the network ingress into the
residence - the home gateway device.
JDSU: If you think about it, the factors that
allow CPE to be thinner and cheaper are the
same ones than enable T&M to address CPE
monitoring. The virtualisation initiative is
systematically converting CPE functions to
SW modules, commoditising the hardware
component and removing the proprietary
The cloud is a big