Data is King: The Rise of QoS and QoE in the Cloud

Content is still king in a controlled mass media environment, but it is constantly under threat of being commoditized. With multiple delivery platforms, multi-screen devices and growing global competition, we live in an uncontrolled digital wild west that is expanding and spinning on every axis. Navigating this ever changing landscape along with increasingly finicky and demanding viewer behaviors, without invading their privacy and blowing your budgets, is quite a challenge.

These new trends are a reality. Collecting reliable, real-time data from a growing number of sources has become the new goal for success. This requires a lot of data from both traditional and non-traditional sources. Some of this data is reliable and readily available and simply needs to be processed in a variety of different ways. Data collected directly from sources such as end users can be unrelenting, and requires a great deal of statistical analysis to clean up before it can be used. Other data sources are not readily available because it is too difficult or impractical to collect.

How important is quality in this competitive, customer engaging mix? Is the new baseline for quality defined as “good enough” to balance a viewer’s expectations with the costs to deliver to a specific platform? Has this become the new standard or is it an opportunity to differentiate? Quality has proven to be extremely important in every industry, such as the automobile industry, in attracting and maintaining customer loyalty with winners and losers determined by which of them underestimated the importance of quality. Once the seconds of black, freeze, blockiness and silence on a typical program are calculated, one could argue QoS and QoE is worse today than it was 10 years ago.

Important Considerations

Digital media and their associated delivery platforms have become very dynamic as technology related to efficiently delivering content continues to evolve. This new technology brings with it new problems and requires new solutions to address these problems. Some important considerations before determining the most appropriate approach are as follows:

  1. What to monitor (either Equipment and Infrastructure or Streams and Content)?

Current systems such as Network Management Systems (NMS) contain system level data collection, aggregation and a level of analysis, but focus on equipment health. They remain useful but unable to completely or efficiently monitor today’s digital media delivery. QoS and QoE systems also contain data collection, aggregation and analysis with focus on Streams and Content. Today’s digital problems require much more of the stream and content monitoring than equipment and infrastructure monitoring, but are very complementary and able to be used together.

  1. How to monitor (On Premise or in the Cloud)?

Many organizations have stringent policies with respect to security and maintaining complete control of all systems. This is a valid approach, but it is costlier without the benefit of the scale, hardware, maintenance, and other system integration associated with the cloud. The cloud pushes all of this responsibility onto the vendor.

  1. Who will monitor (Technology/Skills)?

With all the delivery platforms, new technology, and business models to support, there just aren’t enough resources to cover all bases, much less the right breadth and depth of skills. Monitoring as a Service (MaaS), enabled by the cloud and only offered by Qligent, is an attractive option to address skills shortages such as RF, transport streams, networking, CDNs, etc. For the same reasons as SaaS, MaaS can be accomplished more efficiently than if the same day to day functions were staffed internally, allowing critical staff to work on more strategic activities.

  1. Where to invest (as a service or capital expenditure)?

Some budgeting restrictions only allow for capital expenditures and the associated process of RFI, vendor evaluation, RFP, budgeting, vendor selection, approvals, and acceptance. A SaaS model will allow for a lower cost of entry, reduced time to benefit, rapid prototyping, pay as you go, easy to expand or modify, and if all else fails an easy exit. SaaS will offer the most flexible terms, shares costs and risks.

 Cloud Education

The cloud offers a tremendous value proposition, enabling more streamlined, efficient, reliable, scalable, redundant, secure and cost-effective media workflows. In reality, most have reported it is not really less expensive if compared apples to apples.

The main reason is that most want to work the same exact way as before but with different tools. In this case you will only see an incremental improvement, get frustrated and call it all a bunch of hype. The same was true with the analog to digital conversion. The first step was to replicate the analog service, which was already well optimized. It wasn’t until this new digital technology was exploited that anyone reaped the real fruits.

The same is true for cloud workflows. Slowly, more workflow components will be virtualized and safely operated from the cloud. Once enough of these components are in place, new workflows will start to take shape but they will not look the same as they did before.

Cloud is a new enabling technology, but also a new approach operationally as it moves from the time-consuming and expensive CapEx approval process to a small pay-as-you-go OpEx expense approval. The industry is evolving far too rapidly for the old decision processes to work effectively. By the time you have a CapEx purchase approved, there is already a new technology to replace it. Embrace this reality!

Cloud, SaaS, and related technologies enable rapid prototyping which enables testing and trying in small bits to prove models and ROI with very small investments. Broadcasters can place many small bets instead of a lot of strategic planning, evaluation and budgeting. Concerns about proven technology, security, redundancy and scale are offset by successes in other industries. Education, Marketing, Healthcare, IT, Banking, Travel, Manufacturing and even Government organizations have taken the plunge, and proven the benefits of instilling cloud workflows. It does not need to be built in-house; the broadcaster can partner with trusted players such as Microsoft, Amazon, Google, IBM, Qligent and others. This enables the broadcaster to  share the costs and risks of the major components, and instead focus on how to virtualize components for migration to the cloud. Some are easier than others, and workflows such as monitoring, which depend on specific geographical locations, can be challenging to deploy – but the rewards in efficiency, cost and scalability far outweigh the risks.

Delivery Platform Monitoring

As mentioned, many before us have exploited the cloud and major players such as Microsoft are in direct support of the Media and Entertainment industry. Perhaps it’s time to start trusting that the aggregation, analytics, and integration portion of monitoring can be accomplished in the cloud, and shift our focus on how to virtualize other QoS/QoE workflow components.

First, be cautious not to overbuild. This is easy to avoid with a cloud solution as you can start small and lightweight, find out what you don’t know and expand, modify, and adapt from there.

Also be cautious to keep light weight with your monitoring to avoid trading off real-time feedback for a large quantity of data.

Let’s look at the layers of the stream and why the location of the monitoring is important.

-First the audio and video is produced and processed into the correct format or profile for the intended service and end device.

-Second the streams of audio and video are multiplexed with other streams and data services into transport streams.

-Third they are wrapped and unwrapped as often as required by a variety of physical layer mediums (Baseband, RF, IP, etc.) until they reach their intended audience

The location is important as the number of points grow exponentially as you approach the end user. A good rule of thumb is to keep the cost of monitoring within each ring of the delivery chain flat, meaning you are trading off depth of monitoring for distribution of monitoring. This is not trivial as the parameters monitored are decreasing linearly as the quantities are growing exponentially.

Qligent Info

The best way to accomplish this task is to separate the hardware from the software. This requires software probes, which can run on a wide variety of platforms. Preferably these platforms are already deployed on existing standard servers in rack rooms, as a system on a chip, and on set-top boxes, software players and consumer devices. Another evolving technology trend is how the Internet of Things (IoT) facilitates this architectural approach and drives down costs.

Standard commercially off the shelf (COTS) hardware and popular operating systems (OS) can analyze everything from the transport stream and below, leaving the physical layer monitoring as the major hardware consideration. Take the example of an RF link. In this case you can deploy one such RF probe that does a deep dive on the physical layer in a location representative of a worst case scenario. A low-cost RF receiver approach can be used on all the probes. Correlating and extrapolating the parameters from all these probes is where the true picture is recreated.

It is difficult to predict the future, but the trends, as disruptive as they may be, are real.

One of the best ways to mitigate risk during these uncertain times is with a cloud approach, which can be deployed incrementally. Unfortunately, like any new technology the ROI will come later, and in larger steps once enough workflow components are virtualized and pieced together.

Once a comprehensive deployment of QoS and QoE telemetry is virtualized in the cloud, the true quality of your media delivery will be unveiled. Stimulus and response and/or cause and effect of bit rates, error corrections, buffering, profiles and other configurations can be proven against a wide variety of conditions.

Once this layer is understood, the broadcaster can start overlaying this with other sets of big data in real-time to form very useful results, such as Business Impact Analysis (BIA) and Predictive Analytics. Imagine the possibilities when you combine datasets such as Media Assets; Advertising; Social Media; Ratings; Marketing; Sales; Audience behavior; Major or Current events such as holidays, political elections, sports, concerts, even weather events to create new insights.

One thing has not changed in our industry. At the end of the day all these “new media” distribution models, analytics, audience behavior, etc. will not matter if you fail to deliver a high quality user experience.

Broadcast Beat - Production Industry Resource