Skip to main content

6 Valuable Takeaways from the Internet Fails of 2021

Avoiding Internet failures requires a strategic prevention and preparation plan, solid change management process and a holistic monitoring and observability solution
Mehdi Daoudi

The adjective "disruptive" applied to the year 2021 is an apt descriptor of the tumultuous change we have all experienced due to the many transformations wrought by the pandemic. Digital transformation has delivered a two-edged sword of salvation and chaotic complexity. We have also seen Internet disruptions and outages become more frequent, a worrying trend with an increasingly remote workforce heavily reliant upon distributed cloud-based apps.

Our growing dependence on the cloud and Internet for business means we must take time to prepare for downtime and latency issues. There are valuable lessons found in most failures, and the Internet outages of 2021 certainly provide ample motivation to revamp processes for mitigating system disruptions. Here are six take-aways from 2021's Internet fails that can be used to increase efficiencies in managing the system infrastructure of any enterprise, no matter its size or sector.

1. Even a small change can lead to a major fail

Even the most technically sophisticated business can experience a serious system glitch. Almost all outages are the result of a manual or automated change to code or configuration.

Developing a rigorous change management approach and putting solid protocols in place will help to manage change and its possible consequences. It's important to establish clear policies and procedures around every change, with rollback steps in place for quick restoration when needed.

Your approach should involve tracking every change, testing every change before deployment, and monitoring all services, transactions and outputs that may be impacted if things do go wrong.

2. Monitor beyond your own areas of control

IT typically monitors those areas in which they are most active, like VMs, hardware and code. But they must look beyond an assumption that code bugs or infrastructure load issues are the primary causes of failure. It is equally crucial to observe what is actually delivered to consumers or users.

For an end-to-end view, IT needs visibility into areas outside of their control, such as third-party CDNs, managed DNS, and backbone ISPs. This will allow IT teams to act quickly in the event of a failure, whether that's dropping a third-party, switching to a backup solution, and of course, clearly communicating with users while teams work to resolve the situation.

3. Know the foundations of your network

Many of us live by the old adage, "If it a ain't broke, don't fix it." In the context of system infrastructure, this is often applied as, "If it ain't fixed, it won't break." If there have been no changes or modifications, we often make the mistake of assuming everything is stable.

Unfortunately, this mindset may lead you to miss those single points of failure in your system infrastructure that are rarely changed, such as DNS, BGP, and TCP configurations. All system components need continuous monitoring. Equally, teams must be prepared with a solid plan of action and having regularly practiced their response.

4. I trust you, but let's double check

When another team or vendor is making a change, it is easy to simply trust they have initiated the proper planning and analysis to make sure it's a success, but it's essential you take your own measures to verify this since the outcomes are so crucial.

Using a "Trust and Verify" approach ensures that all the checks and balances are in place when determining the impact of a change. It is essential to have a crisis call plan in place that outlines who is on call, what to do, and who to notify about the specific issue.

Other essentials are a mitigation plan for the failure, which has been pre-tested, and a communication process with templates that include need-to-know info for users and customers. Moreover, developing a monitoring and observability plan is crucial for covering all aspects of system analysis and awareness.

5. An experience monitoring and observability platform solution is your fail-safe

Deploying a holistic monitoring and observability solution platform, that enables deep visibility into all internal system components and the entire delivery chain, should be an enterprise essential. This ensures independent monitoring of every potential point of failure, which ensures you can detect outages and issues from anywhere in real time.

By establishing a baseline for how things look before a change is made, you can understand the impact of the change in regards to areas such as latency, dropped connections, slower DNS servers, and so on. As opposed to simply looking at code tracings and logs, there should be continual testing and evaluation of the output of IT services from the perspective of the end user. Monitoring must be conducted both inside the product environment and outside for 360 degree visibility into the experience.

It is also important not to rely on a cloud-only monitoring and observability solution, which can leave dangerous blind spots across the service delivery chain, and inaccurately report the end user experience.

DNS observability is essential, for instance, since a DNS problem can cause havoc and lengthy outages, like the one experienced on October 1, 2021 at Slack, one of the world's largest collaboration and messaging apps where the core problem was due to a DNS misconfiguration. Users in need of Slack's services were unable to access the app, nor or did they know why since the Slack status page was also down. The outage lasted over 15 hours, as the teams at Slack tried to discern the root cause.

If an enterprise has a monitoring plan and solution that includes a combination of observation across backbone and last mile networks, they would be able to collect data on the availability and performance of real end users trying to access digital services on their home or office networks, including pinpointing DNS as the root cause.

6. Ultimately, communication is key

A communication plan that lays out responsibilities for every role and clear contact channels can alleviate confusion during an outage. It is important to take control of media communications with clearly laid-out protocols, including the potential involvement of a PR firm in the case of a substantial Internet outage, to prevent your enterprise from being a victim of speculation and brand erosion.

Take a lesson from the November 16 Google Cloud outage and establish a process where you are able to change the DNS or CDN configuration to point users to a clearly designed error page that acknowledges the failure and assures resolution with honesty and transparency. Practice your communication plan regularly, so that teams are always prepared, and your end users know what is going on.

The Latest

E-commerce is set to skyrocket with a 9% rise over the next few years ... To thrive in this competitive environment, retailers must identify digital resilience as their top priority. In a world where savvy shoppers expect 24/7 access to online deals and experiences, any unexpected downtime to digital services can lead to significant financial losses, damage to brand reputation, abandoned carts with designer shoes, and additional issues ...

Efficiency is a highly-desirable objective in business ... We're seeing this scenario play out in enterprises around the world as they continue to struggle with infrastructures and remote work models with an eye toward operational efficiencies. In contrast to that goal, a recent Broadcom survey of global IT and network professionals found widespread adoption of these strategies is making the network more complex and hampering observability, leading to uptime, performance and security issues. Let's look more closely at these challenges ...

Image
Broadcom

The 2025 Catchpoint SRE Report dives into the forces transforming the SRE landscape, exploring both the challenges and opportunities ahead. Let's break down the key findings and what they mean for SRE professionals and the businesses relying on them ...

Image
Catchpoint

The pressure on IT teams has never been greater. As data environments grow increasingly complex, resource shortages are emerging as a major obstacle for IT leaders striving to meet the demands of modern infrastructure management ... According to DataStrike's newly released 2025 Data Infrastructure Survey Report, more than half (54%) of IT leaders cite resource limitations as a top challenge, highlighting a growing trend toward outsourcing as a solution ...

Image
Datastrike

Gartner revealed its top strategic predictions for 2025 and beyond. Gartner's top predictions explore how generative AI (GenAI) is affecting areas where most would assume only humans can have lasting impact ...

The adoption of artificial intelligence (AI) is accelerating across the telecoms industry, with 88% of fixed broadband service providers now investigating or trialing AI automation to enhance their fixed broadband services, according to new research from Incognito Software Systems and Omdia ...

 

AWS is a cloud-based computing platform known for its reliability, scalability, and flexibility. However, as helpful as its comprehensive infrastructure is, disparate elements and numerous siloed components make it difficult for admins to visualize the cloud performance in detail. It requires meticulous monitoring techniques and deep visibility to understand cloud performance and analyze operational efficiency in detail to ensure seamless cloud operations ...

Imagine a future where software, once a complex obstacle, becomes a natural extension of daily workflow — an intuitive, seamless experience that maximizes productivity and efficiency. This future is no longer a distant vision but a reality being crafted by the transformative power of Artificial Intelligence ...

Enterprise data sprawl already challenges companies' ability to protect and back up their data. Much of this information is never fully secured, leaving organizations vulnerable. Now, as GenAI platforms emerge as yet another environment where enterprise data is consumed, transformed, and created, this fragmentation is set to intensify ...

Image
Crashplan

OpenTelemetry (OTel) has revolutionized the way we approach observability by standardizing the collection of telemetry data ... Here are five myths — and truths — to help elevate your OTel integration by harnessing the untapped power of logs ...

6 Valuable Takeaways from the Internet Fails of 2021

Avoiding Internet failures requires a strategic prevention and preparation plan, solid change management process and a holistic monitoring and observability solution
Mehdi Daoudi

The adjective "disruptive" applied to the year 2021 is an apt descriptor of the tumultuous change we have all experienced due to the many transformations wrought by the pandemic. Digital transformation has delivered a two-edged sword of salvation and chaotic complexity. We have also seen Internet disruptions and outages become more frequent, a worrying trend with an increasingly remote workforce heavily reliant upon distributed cloud-based apps.

Our growing dependence on the cloud and Internet for business means we must take time to prepare for downtime and latency issues. There are valuable lessons found in most failures, and the Internet outages of 2021 certainly provide ample motivation to revamp processes for mitigating system disruptions. Here are six take-aways from 2021's Internet fails that can be used to increase efficiencies in managing the system infrastructure of any enterprise, no matter its size or sector.

1. Even a small change can lead to a major fail

Even the most technically sophisticated business can experience a serious system glitch. Almost all outages are the result of a manual or automated change to code or configuration.

Developing a rigorous change management approach and putting solid protocols in place will help to manage change and its possible consequences. It's important to establish clear policies and procedures around every change, with rollback steps in place for quick restoration when needed.

Your approach should involve tracking every change, testing every change before deployment, and monitoring all services, transactions and outputs that may be impacted if things do go wrong.

2. Monitor beyond your own areas of control

IT typically monitors those areas in which they are most active, like VMs, hardware and code. But they must look beyond an assumption that code bugs or infrastructure load issues are the primary causes of failure. It is equally crucial to observe what is actually delivered to consumers or users.

For an end-to-end view, IT needs visibility into areas outside of their control, such as third-party CDNs, managed DNS, and backbone ISPs. This will allow IT teams to act quickly in the event of a failure, whether that's dropping a third-party, switching to a backup solution, and of course, clearly communicating with users while teams work to resolve the situation.

3. Know the foundations of your network

Many of us live by the old adage, "If it a ain't broke, don't fix it." In the context of system infrastructure, this is often applied as, "If it ain't fixed, it won't break." If there have been no changes or modifications, we often make the mistake of assuming everything is stable.

Unfortunately, this mindset may lead you to miss those single points of failure in your system infrastructure that are rarely changed, such as DNS, BGP, and TCP configurations. All system components need continuous monitoring. Equally, teams must be prepared with a solid plan of action and having regularly practiced their response.

4. I trust you, but let's double check

When another team or vendor is making a change, it is easy to simply trust they have initiated the proper planning and analysis to make sure it's a success, but it's essential you take your own measures to verify this since the outcomes are so crucial.

Using a "Trust and Verify" approach ensures that all the checks and balances are in place when determining the impact of a change. It is essential to have a crisis call plan in place that outlines who is on call, what to do, and who to notify about the specific issue.

Other essentials are a mitigation plan for the failure, which has been pre-tested, and a communication process with templates that include need-to-know info for users and customers. Moreover, developing a monitoring and observability plan is crucial for covering all aspects of system analysis and awareness.

5. An experience monitoring and observability platform solution is your fail-safe

Deploying a holistic monitoring and observability solution platform, that enables deep visibility into all internal system components and the entire delivery chain, should be an enterprise essential. This ensures independent monitoring of every potential point of failure, which ensures you can detect outages and issues from anywhere in real time.

By establishing a baseline for how things look before a change is made, you can understand the impact of the change in regards to areas such as latency, dropped connections, slower DNS servers, and so on. As opposed to simply looking at code tracings and logs, there should be continual testing and evaluation of the output of IT services from the perspective of the end user. Monitoring must be conducted both inside the product environment and outside for 360 degree visibility into the experience.

It is also important not to rely on a cloud-only monitoring and observability solution, which can leave dangerous blind spots across the service delivery chain, and inaccurately report the end user experience.

DNS observability is essential, for instance, since a DNS problem can cause havoc and lengthy outages, like the one experienced on October 1, 2021 at Slack, one of the world's largest collaboration and messaging apps where the core problem was due to a DNS misconfiguration. Users in need of Slack's services were unable to access the app, nor or did they know why since the Slack status page was also down. The outage lasted over 15 hours, as the teams at Slack tried to discern the root cause.

If an enterprise has a monitoring plan and solution that includes a combination of observation across backbone and last mile networks, they would be able to collect data on the availability and performance of real end users trying to access digital services on their home or office networks, including pinpointing DNS as the root cause.

6. Ultimately, communication is key

A communication plan that lays out responsibilities for every role and clear contact channels can alleviate confusion during an outage. It is important to take control of media communications with clearly laid-out protocols, including the potential involvement of a PR firm in the case of a substantial Internet outage, to prevent your enterprise from being a victim of speculation and brand erosion.

Take a lesson from the November 16 Google Cloud outage and establish a process where you are able to change the DNS or CDN configuration to point users to a clearly designed error page that acknowledges the failure and assures resolution with honesty and transparency. Practice your communication plan regularly, so that teams are always prepared, and your end users know what is going on.

The Latest

E-commerce is set to skyrocket with a 9% rise over the next few years ... To thrive in this competitive environment, retailers must identify digital resilience as their top priority. In a world where savvy shoppers expect 24/7 access to online deals and experiences, any unexpected downtime to digital services can lead to significant financial losses, damage to brand reputation, abandoned carts with designer shoes, and additional issues ...

Efficiency is a highly-desirable objective in business ... We're seeing this scenario play out in enterprises around the world as they continue to struggle with infrastructures and remote work models with an eye toward operational efficiencies. In contrast to that goal, a recent Broadcom survey of global IT and network professionals found widespread adoption of these strategies is making the network more complex and hampering observability, leading to uptime, performance and security issues. Let's look more closely at these challenges ...

Image
Broadcom

The 2025 Catchpoint SRE Report dives into the forces transforming the SRE landscape, exploring both the challenges and opportunities ahead. Let's break down the key findings and what they mean for SRE professionals and the businesses relying on them ...

Image
Catchpoint

The pressure on IT teams has never been greater. As data environments grow increasingly complex, resource shortages are emerging as a major obstacle for IT leaders striving to meet the demands of modern infrastructure management ... According to DataStrike's newly released 2025 Data Infrastructure Survey Report, more than half (54%) of IT leaders cite resource limitations as a top challenge, highlighting a growing trend toward outsourcing as a solution ...

Image
Datastrike

Gartner revealed its top strategic predictions for 2025 and beyond. Gartner's top predictions explore how generative AI (GenAI) is affecting areas where most would assume only humans can have lasting impact ...

The adoption of artificial intelligence (AI) is accelerating across the telecoms industry, with 88% of fixed broadband service providers now investigating or trialing AI automation to enhance their fixed broadband services, according to new research from Incognito Software Systems and Omdia ...

 

AWS is a cloud-based computing platform known for its reliability, scalability, and flexibility. However, as helpful as its comprehensive infrastructure is, disparate elements and numerous siloed components make it difficult for admins to visualize the cloud performance in detail. It requires meticulous monitoring techniques and deep visibility to understand cloud performance and analyze operational efficiency in detail to ensure seamless cloud operations ...

Imagine a future where software, once a complex obstacle, becomes a natural extension of daily workflow — an intuitive, seamless experience that maximizes productivity and efficiency. This future is no longer a distant vision but a reality being crafted by the transformative power of Artificial Intelligence ...

Enterprise data sprawl already challenges companies' ability to protect and back up their data. Much of this information is never fully secured, leaving organizations vulnerable. Now, as GenAI platforms emerge as yet another environment where enterprise data is consumed, transformed, and created, this fragmentation is set to intensify ...

Image
Crashplan

OpenTelemetry (OTel) has revolutionized the way we approach observability by standardizing the collection of telemetry data ... Here are five myths — and truths — to help elevate your OTel integration by harnessing the untapped power of logs ...