Looking back a year from now, this is one annual roundup list you won’t want to make. Throughout 2014, we bore witness to an eclectic assortment of costly web performance hiccups that met the public eye, spanning many different industries, causes and end-users.
Yes, your typical holiday season retail site outage is in the mix, but with this post we wanted to highlight some of the more out-of-the-ordinary examples. You'll see that—while always among the most affected—are just the tip of the iceberg when it comes to the impact of poor application performance as the world becomes more and more connected digitally. So here is a collection of some of the most noteworthy web performance blunders from 2014—particularly in three main industries that seemed to bear more than their fair share of the brunt the past 12 months—and seize the opportunity to reflect on how we might be able to prevent history from repeating itself this year in 2015.
First, we'll start by getting the big, bad holiday season retail example out of our systems as a warm-up before moving on to some of the less typical cases…
Vertical I: Online Retail
The retail industry is obviously the most directly affected financially when their eCommerce site fails or performance degrades. eCommerce sales increased nearly 15% year over year this past holiday season. eRetail sales for two months, just counting purchases made on computers, totaled $53.305 Billion, up from $46.546 Billion a year prior. Clearly, any failure in an eCommerce application during peak times is a very serious matter for a retail business, and especially given these figures. Even so, it seems that every year spit hits the fan for at least one leading online retailer or two at exactly the wrong time—2014 being no exception:
- During the height of holiday shopping season, CNBC reported an especially notable failure that made global headlines:“Best Buy’s Website Suffers Black Friday Outage”
The CNBC.com article, which received over 1,400 social shares and 159 comments, reported that “The company initially attributed the blackout to heavy traffic. However, consumers were still complaining about the lack of service early Saturday, suggesting the problem could be more serious.”
Takeaway: So what’s an eCommerce company to do when even the largest operations with so many resources can fail? In one word, “prepare.” Our recommendations are that you monitor your site all year long – looking at all your critical transactions, 24x7, from locations all over the world. It’s important to test and monitor your applications using the best possible user emulations, testing from inside the browser through the entire application delivery chain and back. This allows you to understand what “normal” is, and how traffic affects your performance. Load test any changes to the application to compare performance metrics against your established baseline, monitor the applications in pre-production, and then use the same test scripts once deployed. This story points out the imperative to load test, monitor in pre-production, and have the right tools to find the problems as fast as possible if you miss something.
Vertical II: Government
As hard and as often as they try to implement digital solutions, Government agencies continue to flounder. Last year, we saw several such government organizations follow in the footsteps of Healthcare.gov:
- The state of Georgia voter information website crashed. On Election Day. The alternatives for finding voter information were not good. The online media site, Fresh Loaf, covered the story, reporting that “…Brandon Hanick of left-leaning political group, Better Georgia, raised concerns about the proposed alternatives. In a statement, Hanick said he tried calling the number this morning, but said he received a busy signal.” Shedding further light, also added "Clearly, there's [no] way the office could handle the volume of voter information requests by phone," he said." I left a message with the Elections Dept. over an hour ago and have not received a return phone call."
- In June 2014, the FCC website crashed after comedian and HBO host, John Oliver, encouraged his viewers to share comments about the agency online. Read all about how not funny this website crash was for the FCC in this article published on theguardian.com that received over 22,700 social shares and 133 comments (which is not all THAT bad, when compared to the video which received more than 7.5 MILLION views on YouTube!) Ouch.
- The Massachusetts Registry of Motor Vehicles also made the news for an unfortunate website crash in December 2014, as reported by the Boston Globe and Firstpost.com (among others, of course), with the latter commenting, “The DDA today launched its much-awaited 'Housing Scheme 2014' offering over 25,000 flats across various categories, amid a huge response from people. But somewhat on expected lines, the authority's website crashed soon after the scheme was opened to the public.”
Takeaway: In the case of the government, it just seems to be a struggle to get the right work done and tested before they draw people to the sites, even if they expect a massive rush. I suspect part of this is lack of understanding of the requirements for making sure that the applications work in real life, and perhaps also not leaving the time to do that work. Of course, as we see in the case of the FCC crash for example, it’s not always clear that random, “viral” events can be predicted. But they can be prevented— if you are prepared with the right monitoring and troubleshooting tools, that is. Digital transformation has made it possible so that any individual with a WIFI connection has the power to influence online traffic and behavior at a mass level at any given time, which means your application needs to be ready for anything, at all times.
Vertical III: Technology
Ah, the wonderful world of technology. One of the more interesting things about application failures in this sector is that one can make the argument that organizations that fall under the tech umbrella are inherently expected to be—by nature— you know, technologically advanced and stuff. Thus, it can be all the more surprising when a tech giant’s application goes down—like if a Gatorade factory were to shut down unexpectedly due to employee dehydration—“isn’t that, like, your whole big thing?”
- In a single week in September 2014, eBay, PayPal, FaceBook, Tumblr, LinkedIn, and Apple all either had significant performance issues or saw their sites go down completely. TechRepublic.com posted an interesting article covering this, which also accuses many of these tech giants as being repeat—in some cases even serial—offenders for application failures throughout the year.
- In 2014 Salesforce.com was named a leader in the CRM Gartner Magic Quadrant for the fourth consecutive year, and their application upon which thousands upon thousands of businesses run (especially their own) went down unexpectedly in April, and then again in September.
Takeaway: One lesson we can learn from the Salesforce.com example is that it’s important to monitor all of your critical business processes, including external, third-party apps. With your own information about third party performance or outages, you have greater control over the issues, and will have time to apply damage control to minimize the impact on your business when problems arise. The right performance monitoring solution can tell you whether the problem resides in your end of the application, or the vendors’. If your own application is the culprit, you’ll have the necessary information to deploy a swift and timely fix, and if the vendor is to blame, you’ll be ready with objective performance data as evidence for any SLA agreements that may have been violated.
It happens to the best of us, that much is clear. But when sites like FaceBook, PayPal or Salesforce.com stumble, so do thousands—sometimes millions—of others that rely on these applications to always be working. This point is a significant one, for as our lives and businesses continue to become more and more tightly bound by the interconnections of technology, the more important it is that those connections remain strong and stable at all times, and the more we the users, expect perfection of them.
You might also be interested in: