Achieving Peak Streaming Performance
Through Video Delivery Optimization
As video streaming matures into mainstream content
viewing, top publishers are challenged to maintain high QoE
for every viewer on every device. Network conditions are
constantly changing, with spiky peaks and troughs based on
traffic, CDN, or net work access provider issues. Increasingly,
large online broadcasters are turning to a multi-CDN strategy
to add a level of redundancy and to reduce issues such as
buffering, downshifts, and interruptions.
Getting the most out of a multi-CDN workflow requires
accurate intelligence that can detect and analyze changing
network conditions, predict network hazards, and make
intelligent, performance-based routing decisions in real-time.
BETTER DATA, BETTER DECISIONING
DLVR (pronounced “deliver”) employs real-time
measurement of every stream to feed data to predictive
algorithms that in turn drive the intelligent decision-making
behind the powerful, multi-CDN management solution.
Here’s how it works:
1. DLVR measures CDN performance for each stream, from
start to finish—how your CDNs perform delivering video
(not small test objects) to your users.
2. DLVR also measures the user’s access network conditions.
If a video buffers, downshifts, or streams at a low bitrate,
DLVR can determine whether that’s the result of the
CDN’s performance or the user’s access network. This is
particularly important for mobile delivery, where a user’s
connection is often the limiting factor.
3. DLVR’s CDN performance measurements enable DLVR
to then predict a CDN’s performance. DLVR doesn’t just
respond to current CDN performance issues, DLVR
anticipates performance issues that haven’t happened yet.
4. Based on these real-time CDN performance measurements
and near-term predictions, plus a customer’s business rules
and priorities, DLVR selects the best CDN for each user on
a stream-by-stream basis, taking into account the user’s
device, access net work, location in the access network, and
the video requested. That means the highest performance
delivery path for each user, every time.
5. DLVR also builds in a next-best delivery path for each user,
making streams instantly resilient in the event of a CDN
blackout or brownout.
Unlike client-side solutions, DLVR runs 100% in the cloud,
requiring no client-side code, SDKs, or Java Script development.
DLVR is implemented in much the same way as a CDN, and
takes about the same time and effort.
A recent CDN failure highlights the value of an effective
multi-CDN strategy using DLVR’s optimization solution:
In late 2017, one of the major CDNs had just rolled out a
software update network-wide containing a defect that caused
content length mismatch errors and client aborts. The moment
the change rolled out, all ongoing streams abruptly stopped
and all new streams immediately failed to start—except for
those from DLVR customers.
Through DLVR’s midstream, client-initiated stream
switching, DLVR’s customers were able to avoid what other
customers of the CDN experienced—a total loss of all viewers.
DLVR’s Responsive Manifest™ designates the CDN selected
to power the video stream, as well as an alternative CDN (the
next-best CDN) to be used if for any reason the first CDN
becomes unresponsive. DLVR can configure any number of
alternative CDNs—as long as at least one of the customer’s
CDNs is available, videos will stream to users.
In this case, 100% of DLVR-optimized streams switched
immediately to an alternative CDN in just 20 seconds,
worldwide, and with no noticeable increase in buffering or
other loss of user QoE: