DEMAND SURGES: CAN
YOUR CDNS KEEP UP?
In October 2017, a large
European broadcaster launched
the season premiere of a popular
television series without advance
notice to DLVR or to its CDNs.
DLVR continually monitored,
measured, and optimized CDN
performance in association with
end-user access networks for peak
video playback. As a 10x surge of
viewers rapidly came online at the
start of the show, DLVR selected
the best delivery path for each user
in real time, factoring in constantly
changing conditions and ensuring
the best possible viewing experience for the entire audience.
The charts from the DLVR dashboard illustrate how the
most performant CDN changed before, during, and after the
premiere, and how DLVR detected, analyzed, and predicted the
best route for each individual viewer. This level of insight and
optimization has not been available to publishers until now.
During the premiere, DLVR’s per-stream CDN selection
algorithm began predominantly choosing CDN 1 for both
Predicted Performance and Hazard Avoidance-based decisions.
After the premiere, CDN 2 was selected 74% of the time,
while CDN 3 received the remaining 26%. CDN 1 no longer
received any traffic based on predictive or hazard-based
CDNs still do the heavy lifting of video stream delivery,
but as video streaming hits mainstream adoption, delivery
optimization services such as DLVR play a critical role
in each video request. Over the course of our first year of
production services, we optimized the delivery of 4. 3 billion
individual video streams to users throughout the Americas
and Europe. That’s 4. 3 billion times that a user pushed
“play” and DLVR was called into action, measuring and
analyzing each of the 4. 3 billion streams to detect, predict,
and mitigate video streaming catastrophes, large and small.
DLVR managed this volume of service while maintaining
99.998% uptime. We are able to achieve this high level of
service through the unique architecture of our video delivery
optimization infrastructure, including the following:
• 100% availability of our geographically distributed,
highly atomic, and highly redundant server and network
• An infrastructure management layer that selects servers and
network paths on a stream-by-stream basis, based on real-time stream performance measurements, instead of using
legacy tools such as DNS-based server selection and BGP
routing for path selection.
• An application software layer designed to handle both
the expected and the unexpected, engineered for high
performance, deployed continuously, and optimized for
For more information on how to deploy DLVR to achieve
peak video streaming performance or to speak to one of our
video delivery experts visit www.DLVR.com.
© Copyright, DLVR 2017
DLVR (pronounced “deliver”) provides advanced technology
solutions to improve internet video viewing experiences for
every user, on every device. DLVR’s video delivery optimization
service ensures streaming video achieves peak performance
and avoids catastrophes large and small. DLVR’s patented
technology uses an industry-first approach that measures CDN
performance across a publisher’s actual streams rather than
via test objects, and optimizes user video requests stream-by-stream, taking into consideration all the factors that influence
the Quality of Experience (QoE) of Internet video for end users.
DLVR works with your existing digital video workflows without
requiring any modifications to apps or CDNs.
DLVR’s purpose-built, video optimization infrastructure and
services are used by leading publishers around the globe to
maximize their multi-CDN strategies and overall quality of
experience. For more information visit www.DLVR.com.
During the premiere, DLVR’s per-stream CDN selection algorithm began predominantly choosing CDN 1 for both Predicted Performance and
Hazard Avoidance-based decisions.
After the premiere, CDN 2 was selected 74% of the time, while CDN 3 received the remaining 26%. CDN 1 no longer received any traffic based
on predictive or hazard-based decisions.