Guess what Target, Neiman Marcus and PayPal shared on Black Friday?
I am sure most of you did not wake up at 4am on Black Friday and stand in the queues. Instead you just had to log on your computer or mobile device in your PJs at 9am. But, did all websites live upto the Black Friday expectations? I am afraid not.
One of the websites message was:
“So sorry but high traffic’s causing delays. If you wouldn’t mind holding, we’ll refresh automatically & get things going ASAP. Thank you for your patience!"
According to NRF reports 103 million Americans shopped online over Thanksgiving through Black Friday weekend, more than the 102 million who shopped in stores.But looks like no one anticipated this consumer behavior. Wonder why!
Neiman Marcus’ website experienced a major outage, which caused it to miss Black Friday altogether. The company responded by having a “Black Saturday” sale instead, but still faced intermittent outage through Sunday morning, only stabilizing around 5:30 AM ET. Other sites like Newegg, HP, new ecommerce site Jet.com, Saks, Victoria’s Secret and even PayPal experienced outages ranging from a few minutes to a few hours.
Now, let us analyze the cost of loss of business due to an outage. Imagine that you have a million customers and the cost of acquisition of a new customer is $150, if 0.5% of them moved elsewhere, it is a $750,000 cost to the business. This does not include the cost of fixing these performance issues and the cost of lost productivity as well as loss of brand image.
What exactly causes these outages? The obvious reason is the volume of traffic and load on the servers. This includes their website, mobile and if there is a wearable app, all traffic bundled into one. But the underlying reason can be quite complex – that can include the design of the website itself, the way data is accessed and the actual capacity of the servers.
So, how do you prepare so that this nightmarish experience does not repeat? Let us look at 3 ways to plan, prepare and prevent an outage during a major event when traffic is expected to spike.
Scalable Cloud Solution: If you do not have a cloud solution, now is the time to think about it. In contrast to a cloud solution, if you are bound by a data center, it is possible to scale. Except that is manual, tedious and not as cost-effective. Services like Amazon Web Services, Rackspace Cloud and Windows Azure offer scalable cloud hosting solutions that can easily be scaled up to get more resources if there’s an unexpected surge in traffic (or scaled down if traffic is low). And you only pay for what is utilized.
Performance testing: Utilizing tools, services and infrastructure perform load testing with simulated real-world traffic patterns. Doing this in a pre-production environment would let you analyze the responses results, based on which you can optimize your applications. Ensure that you always plan in alignment with business needs. Use performance monitoring to stay on top of the issues 24/7.Running a series of tests is key in preparing for such events.Code/ automated releases: As a result of performance testing, you might identify a number of issues that can be code-related. This might be website designs consuming bandwidth. It might also lie in not making database calls efficiently. Sometimes it means that you have not utilized splitting your humungous web applications into independent services. All these code related issues must be caught early on during the testing cycles and fixed and released as quickly as possible. This can be achieved utilizing Agile and automated-code deployment methods.
Conclusion : In conclusion, it is possible to prepare for an event where you expect peak traffic to hit your website (which is great for your business and revenue). It involves careful planning around capacity (Cloud solutions is perfect for this), performance testing and quick code fixes and deployment to enable continuous availability of your web and mobile sites.