Why Monitoring is Essential for Multi-CDN Switching

Many streaming video providers use a multi-CDN strategy to ensure reliable content delivery, improve customer QoE (quality of experience), and reduce the costs of delivering content. As of 2020, 75% of video providers were already multi-CDN or had plans to be, according to a survey by NPAW.

The value of a multi-CDN strategy extends far beyond streaming and video providers. Enterprises must deliver a high QoE around the globe, whether that’s for global brand awareness, the biggest sale of the year, or major gaming releases.

A successful multi-CDN strategy often uses CDN switching, also known as traffic steering. CDN switching involves switching between CDNs based on factors like end user quality of experience, cost, and network traffic. An intelligent approach to switching is only possible with the right data, which is why monitoring is critical to success.

This post will cover:

• The basics of CDN switching.
• Why monitoring is critical for CDN switching.
• The benefits of multi-CDN monitoring beyond CDN switching.
• Why Hydrolix and solutions powered by Hydrolix such as TrafficPeak, an Akamai Cloud observability solution, are the right choice for multi-CDN monitoring and implementing a CDN-switching strategy.

The Basics of CDN Switching

Modern multi-CDN strategies can be either server-side or client-side. With server-side switching, the server determines which CDN will serve requests. Meanwhile, with client-side switching, it’s typically a video player that determines which CDN to make requests to.

How do enterprises choose which CDN should deliver content? Enterprises often use complex business logic to make that determination, usually based on static rules or performance data.

Static rules don’t take performance and QoE into account. Instead, they are stable rules based on factors like ISP, geo, and device type. For instance, customers using a specific mobile device might get content from the same CDN regardless of performance.

Switching can also be based on performance data, which is discussed further in the next section. This typically involves comparing the performance of multiple CDNs in real time, then serving content from the most performant option. The performance data can come from in-house monitoring or third party analytics.

Why Monitoring is Critical for CDN Switching

Multi-CDN monitoring is essential for CDN switching, understanding the overall health of your network, and even other business intelligence use cases for advertising, marketing, and more.

Let’s start by looking at the importance of monitoring for CDN switching.

Switching Based on Performance

Performance-based switching involves dynamically choosing a CDN in real time based on its performance. This can be based on data from your end users or third party data from other providers.

That means monitoring CDNs or using a third party.

For major enterprises, relying on third-party data for performance-based switching isn’t the best approach. Third-party data tends to be a blunt tool, not a precision instrument. For example, you can use this data to get a general aggregate of CDN performance by country or geo. However, that might not be an accurate reflection of QoE or performance for your users.

Instead, it’s much better to make decisions based on your own data, which requires comprehensive monitoring. By doing so, you can more precisely target end users and improve their QoE. For instance, you might want to evaluate performance based on a number of factors.

That could mean a video player queries your CDN data by geo, device type, and ISP to retrieve results such as bitrate and start-up latency. Then the player can input those results into a custom algorithm that determines which CDN to use.

Data based on monitoring end user experience is much more accurate than third-party data.

Switching Based on Static Rules

CDN switching can also use static rules, which aren’t based directly on real-time performance. An example is choosing a CDN based on an end user’s geographic region.

While static rules don’t consider real-time performance, they can and should still draw on historical performance data. In other words, in addition to monitoring CDNs, you should retain the data long-term if possible. In the case of CDN switching based on specific geos, you should be able view CDN performance over time by geo to determine which CDN is the best fit for a static rule.

You can also use historical data to optimize static rules for costs. For example, if you’re seeing less usage in a specific geo, you could use a standard instead of premium CDN tier. You can also have less redundancy for that geo, resulting in lower costs.

Overall System Health and Manual Switching

In some situations such as large real-time events and major outages, enterprises may also incorporate manual switching as well. As an example, if an SRE team is monitoring event dashboards and sees unusual latency in a specific region, they can manually update switching rules.

To do so, teams need to have visibility into system health from the origin to the edge. That way, they can quickly identify issues along with the underlying root cause.

If users in a specific geo are experiencing a lower bitrate or slow start times for video content, monitoring allows you to investigate further. You can then determine if it’s an issue with the CDN, the content, or even a code update.

The Benefits of Multi-CDN Monitoring Beyond CDN Switching

Multi-CDN monitoring can also have benefits that extend far beyond CDN switching, especially if you retain CDN data long-term. CDN data provides deep insights into when, how, and where content is consumed. You can see which assets are popular in each geo and then use that information for advertising, marketing, and business intelligence.

CDN monitoring can also be extended to provide real user monitoring, especially for use cases like streaming video. You can correlate content usage with IP addresses to get a deeper understanding of your customers for customized advertising and recommendations. And you can correlate that CDN data with data from other sources, giving you a comprehensive profile on user devices, video player analytics such as when a user abandoned content, and more.

CDN data contains a wealth of information on user behavior. The ability to analyze user behavior by geo and other dimensions is extremely powerful for BI and data science use cases.

Why Hydrolix for Multi-CDN Monitoring?

CDNs can generate terabytes of data every day for major enterprises. Due to the scale, most observability platforms can’t offer comprehensive CDN monitoring because of the cost of ingesting, storing, and analyzing all that data.

As a result, enterprises have traditionally had challenges with effective CDN monitoring. They might not do it at all, and if they do, they’re typically forced to sample data or keep it available for only a short period of time.

Hydrolix is a real-time data platform for high volumes of log data, including data from CDNs. It provides both real-time analytics and cost-effective long-term storage for historical analytics.

Hydrolix also powers TrafficPeak, an Akamai cloud observability solution. Hundreds of the biggest enterprises in the world currently use TrafficPeak for a wide range of use cases including multi-CDN monitoring. 

Let’s take a look at the reason many enterprises turn to Hydrolix and TrafficPeak for multi-CDN monitoring.

Real-Time Querying for CDN Switching and Analytics

Hydrolix uses techniques like partition pruning and micro-indexing to provide subsecond “needle in a haystack” query latency even on trillion-row datasets. In addition to helping teams troubleshoot issues quickly, you can use Hydrolix in dynamic CDN switching business logic.

With Hydrolix, you can also use summary tables to provide efficient aggregations of CDN data. Summary tables even account for late-arriving data, ensuring that they remain accurate.

Ingest Data from Multiple CDNs Into a Single Table

CDN switching is most effective when business logic can effectively compare data from multiple CDNs. For many platforms, comparisons across different data sources can only happen with compute-intensive JOIN statements that are typically less performant.

Hydrolix has the unique ability to ingest multiple data sources into a single table so that no JOIN statements are needed for comparisons. This is particularly useful for comparing sources that logically belong together like CDNs.

Read more about how Hydrolix can join data from different sources at ingest time, making it much easier to compare data across multiple CDNs.

At ingest time, Hydrolix transforms data based on user-defined specifications, so you can standardize your CDN data across sources. Hydrolix also includes flexible schema, which means that different sources can have different fields. For instance, one CDN might have more granular data available than others. You can still ingest and query that data, and because Hydrolix is a columnar datastore, you can specify just the columns you want to retrieve to minimize compute resources.

Doesn’t that result in very large tables, impacting performance? Hydrolix is designed to efficiently query trillion-row datasets, so you can ingest all your CDN data into a single table for easy comparisons.

This makes it much easier to monitor all your CDNs for multi-CDN switching use cases.

Designed to Handle Large Volumes of Data Cost-Effectively

Cost is one of the biggest barriers to comprehensive multi-CDN monitoring. Ingesting and retaining terabyte volumes of log data every day is simply too expensive for most platforms.

This is not the case for Hydrolix or TrafficPeak. Hydrolix is designed for daily terabyte-scale volumes of data. This volume of data is typical for major enterprises analyzing multiple CDNs. Just as importantly, Hydrolix gives enterprises the most cost-effective per GB cost to ingest, store, and analyze log data in the industry.

See our pricing estimator to estimate your log data costs.

TrafficPeak is also extremely cost-effective compared to other observability solutions.

Easily Scale Up and Down for Major Events and Unexpected Incidents

CDN traffic can be highly variable, leading to spikes in traffic during major events and incidents and lower traffic at off-peak times. To handle this variability, you need a platform that can scale up or down without overprovisioning (leading to excess costs) or underprovisioning (leading to low performance).

Hydrolix uses stateless, cloud-native Kubernetes architecture so that each component can be scaled up or down independently. For example, you can scale ingest components independently from query components. You can even define separate compute pools to prevent resource contention. Hydrolix is designed to autoscale but you can also manually scale to fit your use case. This allows you to rightsize your monitoring architecture to keep costs down while also monitoring large events.

Long-Term “Hot” Retention for Deeper Insights

With Hydrolix, you can retain all your log data cost-effectively for years—and get access to data whether it’s a minute or a year old. This is in stark contrast to platforms that typically offer short hot data retention windows of 90 days or less. Enterprises dealing with terabyte-scale volumes of data on a daily basis are often forced to have even shorter retention windows of 7 days or less to keep costs down.

Hydrolix maximizes the performance of cost-effective, S3-compatible storage to provide long-term data retention. S3-compatible storage is both affordable and highly scalable. Enterprises using Hydrolix can also negotiate discounts with their cloud providers for even greater savings.

With long-term retention, you can get deeper insights into how CDNs are performing over time. Then you can create a proactive multi-CDN switching strategy that anticipates and prevents issues before they arise. This includes the insights you need to define effective static rules for multi-CDN switching. You can apply ML training to help derive insights and detect anomalies. Or you can use the data for other BI and data science use cases.

Correlate Device and Video Player Analytics With CDNs

Monitoring CDNs is just one part of the puzzle. In order to monitor QoE, you need insights where users are actually engaging with content. This includes end user devices such as mobile phones and smart TVs. And considering that video made up 82% of all internet traffic in 2022, it’s also important to monitor the video players where users are engaging with content.

Hydrolix has partnered with Mux to extend observability to the screen, giving you a deeper understanding of end user QoE. You can see how long videos are taking to load, common issues like rebuffering and low bitrates, and exactly when and where users abandon content. You can then correlate these problems with your CDN data to pinpoint where the issue is arising and switch between CDNs as needed.

It’s equally important to have visibility into user devices as well, which is why Hydrolix has partnered with ScientiaMobile. ScientiaMobile can identify more than 100,000 device profiles and provides nuanced data about end user devices.

This way, you can pinpoint if user devices are responsible for issues. For example, let’s say video analytics are showing a poor QoE for a specific geo in India. You might think the issue is CDN-related when it’s actually a compatibility issue with a mobile device commonly used in that geo.

By collecting more data, you can get a better understanding of the root cause of issues and whether CDN switching is the best approach. You can also determine whether end users are having QoE issues that warrant switching between CDNs as well.

TrafficPeak: An Observability Platform for Akamai Users

Akamai is widely considered the top CDN provider in the world. If you’re already using Akamai, you might want to consider TrafficPeak, an Akamai Cloud observability solution powered by Hydrolix. You get all the benefits of Hydrolix including multi-CDN monitoring alongside Akamai data such as SIEM and DS2.

Learn more about TrafficPeak.

Next Steps

Since CDN monitoring is one of many use cases that Hydrolix and TrafficPeak excel at, we’ve written a lot about it. Check out some of our other resources:

• The value of CDN monitoring, complete with visuals and dashboards of valuable metrics.
• Best practices for multi-CDN monitoring.
• How Hydrolix helped a major broadcaster monitor the biggest game of the year.
• And how a major broadcaster implemented a multi-CDN monitoring strategy using Hydrolix.
• If you’re looking for a pre-bundled solution that can help you get started quickly, check out CDN Insights.
• How major enterprises in verticals including retail, banking, and gaming use TrafficPeak for CDN monitoring.