Over the past couple of years, there has been an increasing interest in observability among streaming video operators and how to build it into live stream dashboards. It’s not that traditional monitoring is bad, it’s that observability promises something more: a way to derive insights without doing the manual work of piecing together data points from different sources and streaming formats. Access to the underlying data is critical once insights have been identified, so monitoring and observability must both be available to operations engineers ideally in the same dashboard or tool. Only that’s the challenge: finding a way to make observational insights easy to understand while providing quick access to the data underneath in a single dashboard.
Why the big push for observability?
With traditional monitoring, operations engineers might spend hours connecting data points and finding relationships between data sets to uncover a root cause. Solving performance or quality issues has to be accomplished as quickly as possible to mitigate the negative impact on viewer satisfaction or subscriptions. To get to root causes faster, operations need tools that support observability and can surface connections between data points for faster identification of issues with stream or bitrate availability, network performance, and viewer Quality of Experience (QoE) problems.
Observability, though, is not a silver bullet to lower mean-time-to-diagnose (MTTD). It does not replace monitoring. Effective streaming operations require access to both insights provided through observability and to the underlying data made available through monitoring. Missing one or the other means slowing down the ability to resolve issues quickly. Think about it this way: without observability, operations or network engineers need to spend more time post-processing data and fine-tuning live stream dashboards; while not having monitoring means being unable to see the actual data behind the observations.
It’s safe to say that without both lenses, the streaming experience, and ultimately the bottom line, will suffer.
Live stream dashboard: understanding what monitoring and observability provide
Although there are many definitions of observability (which originally referred solely to IT systems), there is a simple way to understand it. Think of observability as providing an understanding of what’s happening inside a system. Monitoring, on the other hand, provides streaming operators with details about what’s happening. They are different ways to look at the same data: one provides a high-level view while the other provides details. A great way to see the relationship between monitoring and observability is to think about it as a painting. Observability is looking at a painting from a distance to see the entire scene, while monitoring is stepping up to it to examine the individual brushstrokes that constitute the different parts of the picture.
In the streaming world then, observability provides insight and knowledge about what’s happening within the streaming workflow, while monitoring shows the individual data streams from different components.
The problem with streaming operators relying only on monitoring is that they have more data than they know what to do with. Operations are gathering it constantly from a variety of systems. As such, it’s extremely difficult to harness insights from a traditional monitoring dashboard that simply displays data points and metrics.
Although both of these solutions provide certain benefits, the real trouble happens when companies attempt to make one live stream dashboard work for both.
The real problem: one live stream dashboard, two different problems to solve
Streaming observability and streaming monitoring ultimately deliver the same value: providing operations with a way to investigate and improve performance, quality and other issues which undermine the viewer experience. The issue is that different operational audiences need different data. For example, operational leaders require top-level views (observability) to ensure that the streaming service as a whole is healthy. On the other hand, network engineers require data (monitoring) about specific elements of the streaming workflow, such as infrastructure, to determine if the element is operating as it should be.
When both of these audiences are crammed into the same live stream dashboard, the visualisations become far less effective. If the visuals are too high-level, the network engineers will need to turn to another tool to dig into the data. If the visuals are too detailed, operations management will have to spend time post-processing data to understand the holistic health of the streams and the platform.
What does a unified live stream dashboard look like?
It makes sense for the streaming industry to seek a unified dashboard that provides both monitoring data and observability insights. Operations are already using multiple dashboards, so anything that reduces “tool bloat” is readily embraced. The benefits table clearly illustrates how observability and monitoring are different sides of the same coin. So, what does a unified live stream dashboard look like?
First, the dashboard needs to provide a high-level visualisation of the health of the streams. This could be a series of components in a workflow-like design beginning with ingest, moving through encoding, protection, and advertising, and ending with the player. Colours could be used to indicate the health of each component, such as red, yellow and green. This provides the observability operational managers need.
Second, the live stream dashboard needs to provide data points within the visualisation. For example, if a component is marked as yellow and the data points driving that colour are at the low-end of the threshold, then it might not be as high of a priority to investigate as another component where the data points are at the high-end of a yellow or even red threshold.
Regardless of whether the audience seeks observability or monitoring, all the dashboard components should be clickable. This provides engineers and managers with the ability to dig deeper for further investigations and reduce mean-time-to-resolve (MTTR).
A live stream dashboard isn’t just for operations
The real value of a combined observability and monitoring dashboard is that it can apply to more than just operations. If the live stream dashboard collects data from all workflow components, then it can also be used by other streaming teams. For example, product teams can use it to see how features of their applications are performing; business teams can see how advertising is performing: what is the error rate, what are the watch rates; and executives can see the health of subscribers: how much content they are viewing, how long they are spending on the platform, etc.
As streaming operators look to a combined observability and monitoring solution, it’s important to consider all the possible use cases. Reducing tool bloat is critical for better communication and a more unified approach to issue resolution.
The key to live stream dashboards: embrace all the looks of data
Observability and monitoring within video streaming are two aspects of the same challenge: a more in-depth understanding of what’s occurring inside the streaming workflow. Observability provides that high-level insight, while monitoring shows the underlying data. Unfortunately, the visualisation tools used for each of these are often different. Engineers can see the high-level problems in one tool but must jump to a different one, often a separate dashboard, for each data source. All of that increases MTTD and MTTR. Operations need them both.
The key is to combine them in a way that addresses the different audiences within operations (and the company at large): a single live stream dashboard that provides high-level, colour-coded assessment of the streaming workflow’s health yet allows engineers to dig into a connected data source to see what the numbers really mean. Streaming operators who can accomplish this, who can make observability and monitoring co-exist, will ensure their viewers always have the best QoE and the business has the best chance of long-term success.
To find out how you can get the best of both worlds--observability and monitoring--in one dashboard by using a monitoring harness, download our Monitoring Harness White Paper now.