Down Goes the Internet (Again) – Part One: Are You Ready?
October 28, 2013
Klaus Enzenhofer
Share this

What is the cost of downtime? The answer obviously depends on various factors such as the size of an organization, the industry, the duration of the outage and the number of people impacted. To provide a ballpark, though, the Uptime Institute Symposium estimates an average cost of $5,600 per minute.

And it's not all about dollars and cents. Reputation, customer retention, employee satisfaction and overall confidence can be shaken by even a short outage.

For these reasons, media are drawn to outage stories like passers-by of a major roadside accident. Recently, a spate of high-profile outages has once again captured headlines around the world:

- On August 14, the New York Times website experienced a two-hour failure, in which the newspaper had to resort to publishing articles on its Facebook page.

- On the same day, Microsoft customers began reporting email failures en masse. The outage was traced to problems with the Exchange ActiveSync service which serves email to many of the world's smartphones. When Exchange hit a glitch, the volume of phones trying to connect triggered a tsunami of traffic that took three days to get under control.

- On August 22, a software bug and other technology issues brought the NASDAQ stock exchange to a standstill. For approximately three hours, trading was halted for Apple, Google and Facebook and others of this ilk. Because other exchanges rely on NASDAQ's pricing, the fault had a ripple effect that seriously undermined market confidence. This grim fallout resulted in a third fewer shares being traded in the US on that day.

- On August 23, Apple's iCloud service, which helps connect iPhones, iPads and other Apple devices to key services, went down for more than six hours. While Apple claimed that the outage impacted less than one percent of iCloud customers, the sheer size of this user base – 300 million users – translated to approximately three million users being disconnected from services for 11 hours.

- On August 26, an Amazon EC2 outage rocked Instagram, Vine, Netflix and several other major customers of this cloud service, inflicting unplanned downtime across all of them. Last year, a similar Amazon EC2 outage caused Netflix to go down on Christmas day, a busy time for the video streaming service. This occurred in spite of the fact that Amazon had just upgraded their servers to make them less likely to collapse.

In response to these events, industry experts have sounded their alarm and issued a stark warning: we are over-reliant on a digital infrastructure that has become far too complex and exceeds our limits of control.

From high volume securities trading to the explosion in social media and the online consumption of entertainment, the amount of data being carried globally over private networks, such as stock exchanges, and the public internet is placing unprecedented strain on websites and the networks that connect them. According to recent statistics from Cisco, by 2017, the amount of data equivalent to all the films ever produced will be transmitted over the internet in just three minutes.

With these trends showing no signs of abating, we can expect widespread service outages and performance degradations to continue. Knowing this, many organizations go into overdrive as they attempt to improve their resiliency and ensure strong performance levels. But like an auto-immune disease, the addition of processes and technologies can actually have the adverse effect of increasing complexity and risk, by introducing new points of failure.

What is Causing Today's Massive Ripple Effect?

Today, businesses are hosting less and less of what gets delivered on their websites, instead relying on a growing number of externally hosted (third party) web elements to enrich their web properties. These third party internet services are also called web services, and when a major web service goes down, it often takes a portion of the internet with it.

As an example, on August 16, several of Google's websites including email, YouTube and its core search engine suffered a rare four-minute global meltdown. The episode, the cause of which Google has not explained publicly, served to illustrate the staggering volume of global internet traffic served by Google. During the outage, one monitor put the drop in global internet traffic at 40 percent – reinforcing the concept that when major web services fail, they tend to fail spectacularly.

It's certainly true that in recent years, businesses have dramatically increased their use of cloud services. According to Verizon's State of the Enterprise Cloud Report, enterprise use of cloud technology grew by 90 percent between January 2012 and June 2013.

Put another way, to stay competitive, companies have no choice but to provide the best online experience to their online customers and shoppers. That means they must provide the ability to watch a product video, use a coupon, subscribe for a promotion, read customer reviews, share with their friends on Facebook or Twitter, select products and pay for them to be delivered within a defined timeframe. All of this constitutes a lot of minor services. In theory, these functionalities could be developed, hosted and maintained in house, but the cost associated with hardware, software, development, support and maintenance often does not make economic sense.

In the meantime, startups who saw the business opportunity have developed externally hosted packaged solutions for the very functionalities companies wish to offer. The trend is to contract more of these specialized third party service providers, which often results in a company becoming a cloud customer indirectly, without their even knowing it!

Today, a North American website has somewhere between 9 and 13 third party web services contributing to a typical web transaction, according to May 2013 data provided by Compuware's Outage Analyzer. If any one of these third party services slows down or fails, performance for an entire web page, mobile site or application can degrade substantially, wreaking havoc on a company's reputation and revenues.

Third party performance issues occur more frequently than one might think. As an example, Outage Analyzer recently collated data for the six months between March 1 and August 31, 2013 and found 6,217 total outages which included:

- 1,500 full service outages – an average of 125 per month or about four daily – where the entire web service was unavailable in all geographies.

- 4,717 partial service outages – an average of 393 per month or about 13 daily – where only certain geographies or a limited number of user transactions were affected. While full service outages get the most attention, a partial service outage is more likely to occur and affect a limited number of individual web and mobile transactions, while leaving others completely untouched. But all you need is one disgruntled user logging onto Facebook or Twitter to start spreading viral negativity on your brand.

According to Compuware, there are nearly 1,500 distinct third party services available worldwide. Ad servers and social media plug-ins experience the highest number of outages, while online security services and ad verification experience the fewest number. The longest duration for a web service outage in this tracking period was 4,876 minutes (or 3.3 days) for an ad serving firm on March 21, 2013.

Down Goes the Internet (Again) – Part Two: 4 Strategies to Ensure Website Performance

Klaus Enzenhofer is Technology Strategist for Compuware APM’s Center of Excellence.

Share this

The Latest

May 28, 2024
When it comes to ensuring the effectiveness of a software application, it's paramount to ensure that the application can handle varying degrees of demand. Performance testing is a crucial aspect of the software development cycle ...
May 23, 2024

Hybrid cloud architecture is breaking the backs of network engineering and operations teams. These teams are more successful when their companies go all-in with the cloud or stay out of it entirely. When companies maintain hybrid infrastructure, with applications and data residing across data centers and public cloud services, the network team struggles. This insight emerged in the newly published 2024 edition of Enterprise Management Associates' (EMA) Network Management Megatrends research ...

May 22, 2024

As IT practitioners, we often find ourselves fighting fires rather than proactively getting ahead ... Many spend countless hours managing several tools that give them different, fractured views of their own work — which isn't an effective use of time. Balancing daily technical tasks with long-term company goals requires a three-step approach. I'll share these steps and tips for others to do the same ...

May 21, 2024

IT service outages are more than a minor inconvenience. They can cost businesses millions while simultaneously leading to customer dissatisfaction and reputational damage. Moreover, the constant pressure of dealing with fire drills and escalations day and night can take a heavy toll on ITOps teams, leading to increased stress, human error, and burnout ...

May 20, 2024

Amid economic disruption, fintech competition, and other headwinds in recent years, banks have had to quickly adjust to the demands of the market. This adaptation is often reliant on having the right technology infrastructure in place ...

May 17, 2024

In MEAN TIME TO INSIGHT Episode 6, Shamus McGillicuddy, VP of Research, Network Infrastructure and Operations, at EMA discusses network automation ...

May 16, 2024

In the ever-evolving landscape of software development and infrastructure management, observability stands as a crucial pillar. Among its fundamental components lies log collection ... However, traditional methods of log collection have faced challenges, especially in high-volume and dynamic environments. Enter eBPF, a groundbreaking technology ...

May 15, 2024

Businesses are dazzled by the promise of generative AI, as it touts the capability to increase productivity and efficiency, cut costs, and provide competitive advantages. With more and more generative AI options available today, businesses are now investigating how to convert the AI promise into profit. One way businesses are looking to do this is by using AI to improve personalized customer engagement ...

May 14, 2024

In the fast-evolving realm of cloud computing, where innovation collides with fiscal responsibility, the Flexera 2024 State of the Cloud Report illuminates the challenges and triumphs shaping the digital landscape ... At the forefront of this year's findings is the resounding chorus of organizations grappling with cloud costs ...

May 13, 2024

Government agencies are transforming to improve the digital experience for employees and citizens, allowing them to achieve key goals, including unleashing staff productivity, recruiting and retaining talent in the public sector, and delivering on the mission, according to the Global Digital Employee Experience (DEX) Survey from Riverbed ...