Skip navigation

Beware of the Cloud-Only Approach to Monitoring

When enterprises rely solely on cloud monitoring, they often get a skewed, overly positive sense of end users’ digital experiences.

Mehdi Daoudi is CEO of Catchpoint.

Data center and IT operations managers have long understood that monitoring exclusively from behind the firewall won’t tell you if end users are enjoying fast, reliable digital experiences. That’s because there’s a huge canvas of external elements located beyond the firewall – third-party services, ISPs, CDNs, and more – which can impact what end users ultimately experience in the last mile.  

End user monitoring solutions have evolved accordingly, spanning the entire delivery chain. These include a variety of synthetic monitoring options – from the cloud, internet backbone, broadband, ISP, last mile and wireless network locations – as well as real user monitoring. Cloud monitoring, in particular, has gotten a lot of attention. Wouldn’t monitoring from the cloud be the fastest, most efficient way? Absolutely not.

Cloud monitoring has useful applications, but as a solitary solution, it is full of hazards. Here we’ll explore the major drawbacks of this approach, as well as the proper role of cloud monitoring within a more comprehensive performance monitoring strategy.

The Broader Internet Infrastructure

Imagine a foreign dignitary being driven through the streets of New York. He or she has a motorcade clearing a path to enable obstacle-free, fast navigation. Would you estimate the time it takes for an everyday driver to traverse the same route, based on how long it takes the dignitary? Certainly not.

In cloud monitoring, packet traffic is the dignitary. Cloud providers have partnerships with ISPs and better network intelligence on how to route traffic. Whenever possible, cloud monitoring will bypass the broader internet infrastructure with which real end users must contend, keeping packets in transit on their own networks and optimizing speed from point A to point B, and vice versa.

Which brings us to our first key point: When enterprises rely solely on cloud monitoring, they often get a skewed, overly positive sense of end users’ digital experiences, missing the performance potholes across the broader internet. With other types of synthetic monitoring, packet traffic is treated more like real-world, ordinary drivers with no special privileges, traversing the same networks and infrastructures used by the general public.

We recently helped a major e-commerce site identify a slow performing CDN in key geographies (Brazil and India), which gave the organization a head start in working with its CDN partner to resolve the issue. Had this enterprise been using cloud monitoring alone, this pothole would likely have been missed.

Cloud monitoring presents other challenges, such as when an enterprise deploys cloud monitoring nodes to measure the performance of digital services hosted in the same cloud infrastructure. This is like assessing the quality of an international phone call, by testing a call with your next-door neighbor. The local call won’t be exposed to the same choppiness and jitter as the international call, which must pass through many different networks and external systems.

Cloud monitoring applied to adjacent cloud-based services can offer a close-up, clean view of how the service – and the supporting cloud – are performing, but it does not provide an accurate performance view for geographically dispersed end users. Additionally, cloud services are not immune to outages – which should make enterprises think twice before entrusting something as important as performance monitoring to a single cloud service.

Why is Cloud Monitoring Even an Option?

As we’ve discussed, ensuring high quality performance requires data center and IT operations teams to get their arms around increasingly more elements, both within and beyond their firewalls.  As enterprises move more workloads to the cloud, and furthermore, adopt and implement multi-cloud strategies, cloud monitoring does have valid use cases. For example:

  • Performance Monitoring “Lite”: Cloud-based monitoring nodes are available as a shared resource, and can give enterprises a baseline view of digital performance from the vantage point of the cloud.  This is a cost-efficient option for enterprises just starting out with monitoring initiatives, although it by no means represents a comprehensive, fully reliable approach.
  • Private enterprise nodes: Enterprises can deploy a private enterprise node within a dedicated cloud instance. While this doesn’t necessary convey end user performance, it can provide insights and reassurances to the enterprise regarding the overall health of their cloud infrastructure.
  •  Multi-cloud implementations: The trend towards multi-cloud implementations is growing. Consider an e-commerce store with an application front-end residing on one cloud platform, and the transactional shopping cart functionality in another. Ultimately, an end user’s overall strong performance perception – i.e., a speedy, slick conversion – requires ultra-fast, seamless connectivity between the various cloud platforms. Cloud nodes can provide an accurate read on the speed and reliability of interactions between various cloud platforms and regions.

A Comprehensive Approach to Monitoring Is Necessary

When you monitor from multiple vantage points – not just the cloud, but internet backbone, broadband, ISP, last mile and wireless network locations – end user performance degradations as well as their underlying causes are more comprehensively identified, across the complete application delivery chain.

In this context, each individual type of synthetic monitoring data contributes to a complete performance monitoring picture. Each data type provides a unique vantage and correlation point that ultimately makes the whole monitoring dataset inherently richer and more actionable.

Consider the scenario where an enterprise notices their web page load time in a particular region is slowing down, and it needs to figure out why - and quickly. It assesses that cloud-based monitoring data from their regional cloud provider is running slower than usual, while monitoring data from other vantage points is up to par. This should alert the enterprise that this particular cloud service is likely having problems expediting traffic to the website. Armed with this information, the enterprise can narrow their focus and avoid wasting time looking into other elements that are less likely to be the problem.


In many ways, the cloud-only approach is similar to the idea that real-user monitoring (RUM), on its own, can offer a complete monitoring solution. RUM, which measures the behaviors and performance of real end users (actual people) once they enter a website or application, plays a very helpful role in identifying popular landing pages, conversion paths and priority areas for optimization. However it will not alert enterprises to traffic unable to enter the site in the first place. Like RUM, cloud monitoring provides valuable reinforcing insights, but leaves significant performance blind spots.

The cloud may be great for many things, but it is not a sufficient solution for performance monitoring. Enterprises must heed the dangers and not be led to believe they can rely on cloud monitoring alone. In a world of increasing web complexity, more perspectives can only help, but each one on its own cannot and should not be relied upon to paint the whole picture.

Opinions expressed in the article above do not necessarily reflect the opinions of Data Center Knowledge and Informa.

Industry Perspectives is a content channel at Data Center Knowledge highlighting thought leadership in the data center arena. See our guidelines and submission process for information on participating.


Hide comments


  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.