Not all retailers were left unscathed by two of the busiest shopping days of the year, as both high street retailer Game’s online store and The Perfume Shop’s website reportedly crashing with shoppers scrambling to secure Black Friday bargains.
Twitter was awash with customers complaining they could not complete orders and a Perfume Shop statement on Twitter declared ‘we are experiencing a high volume of traffic on the website – we are currently looking into this as a priority and should be up and running shortly.’
>See also: Let’s make it personal: the UK retail digital revolution
On Tuesday of this week, retailers witnessed what some are calling the ‘Meghan Effect’, following the announcement of Prince Harry and Meghan Markle’s engagement. The media interview featured Meghan Markle in a white coat by Canadian fashion firm Line the Label. Once the announcement of the royal engagement was screened, this reportedly set off a shopping frenzy that crashed the Line the Label site.
So how do online retailers work to remove the impact of the Meghan Effect and improve customer’s online shopping experience? It is important to understand the likely cause of the service disruption and to implement best practices that can help to avoid a negative customer experience and ultimately increase revenue.
Identify your weakest link
E-commerce websites are by their very nature complex. They are often distributed across various data centres and clouds, integrating many third-party services, scripts and tags. A positive buyer experience depends on dozens of systems, platforms, services and applications working together. If anyone slows or fails, it could cause a chain reaction affecting the entire experience. The weakest link in the chain can bring the site down.
>See also: To improve your customer experience, free your data
Despite significant advances as we embrace the age of cloud computing, too many retailers continue to place their trust solely on on-premises data centres, with cost compliance and cost control being cited as the most common reasons.
However, by adopting this approach of operating on-premise, organisations are immediately shackled by the set amount of infrastructure available to them. And when demand for IT resources exceeds the data centre’s capacity, the consequences are significant, leading to severe performance issues or worse still, the site to crash.
Infrastructure that can’t meet your demands
When considering infrastructure, it is vital that online retailers have enough capacity in their data centres to meet demand; without it, they risk falling over. Or at least a contingency plan to scale-out or burst into another data centre or cloud. While the argument for connecting multiple data centres and clouds seems logical and straightforward, it is in fact much more complex in practice.
It simply shouldn’t be underestimated, as it can take months of preparation and configuration to create a hybrid environment that maintains consistent visibility, security and control.
>See also: The 3 types of analytics set to transform customer experience
As such, an increasing number of retailers are migrating to public cloud offerings from Amazon, Microsoft and Google for elastic scale to deal with spikes. While transitioning from on-premises to the cloud undoubtedly involves both technological and operational challenges, making the switch can provide improved availability and performance during peak periods.
Old school application delivery controllers
As retailers look to manage a lot of traffic, they need to use a lot of servers. A common approach is for online retailers to use application delivery controllers – often referred to as load balancers – to spread traffic during peak times. While it might sound nonsensical, these load balancers can in fact become the bottleneck during times of peak traffic.
Traditional load balancers are physical appliances that reside in the datacentre, processing SSL transactions and routing traffic to the best available resources. Such traditional load balancers can undoubtedly offload some of the burden from back-end infrastructure, but they too have capacity limits and users should expect outages if traffic comes in at a greater volume or rate than a load balancer can handle. If using hardware load balancers, it is vital to buy enough to satisfy any possible surge in traffic, despite knowing that they will be sat idle during slower periods.
Load balancers are of course still required in cloud environments. This has prompted traditional load-balancing vendors to develop and release virtual editions of their appliances.
>See also: Redefining customer experience: the future of the branch
Simply put, these virtual load balancers maintain a similar architecture to their hardware counterparts, but run on virtual machines. However, as traffic scales in the cloud, online retailers are required to manually configure more of the virtual load balancers to meet the increased level of demand.
Presenting a new alternative to hardware appliance load balancers are software load balancers, running on commodity x86 infrastructure i.e. hardware, virtual machines or containers. The clear and immediate benefit associated with software load balancers is that they provide near real-time elasticity in the data centre and across cloud environments.
Additionally, a software-defined or smart load balancer will include detailed telemetry and analytics about the application, which in turn enables predictive auto-scaling. This ability to automatically bring additional load balancers online provides a system which is much better able to respond to radical traffic fluctuations.
Disruption by third party scripts, tags and services
While traditional load balancers and infrastructure that is not up to scratch can both cause many woes, third-party scripts, tags and services can also be a source of disruption during traffic peaks. Web pages regularly have to load scripts and services hosted by third parties, which might include online ad servers or marketing analytics services. This further complicates the issue of reliability and availability.
A retailer’s web page may well be robust, but its success now also relies on the infrastructure and application delivery solutions of the third party as well. In essence, this creates a complex web of interdependencies.
To keep serving customers and providing the experience they expect, each participant in the complex web must be able to handle the load, no longer simply the one retail site that the customer believes they are buying from.
>See also: How Tesco is using AI to gain customer insight
Whether the peak in demand is due to Black Friday or our soon-to-be princess’s fashion choices, such opportunities are too critical to have poor-performing scripts and tags degrade the customer’s experience.
One positive step is to consolidate your application monitoring and customer tracking solutions to reduce exposure to third-party services failing. Taking it one step further, many modern cloud infrastructure providers and software-defined load balancers now provide detailed analytics that can reduce the need for third party scripts and tags too.
Historically, preparing for a peak in traffic would mean buying and provisioning enough hardware for the peak, which of course then sits idle for the rest of the year. While this is far from cost-effective, for some retailers it was worth the cost.
However, there has to be a better way. Today, with elastic cloud performance, many e-commerce sites can handle their predicted load, without incurring unnecessary costs as they adopt software-based application delivery.
For such savvy businesses, the spoils of success are increasingly great. And with media attention for the impending royal wedding only going in one direction, I’d urge all retailers to ensure they have an elastic software-defined solution able to cope with unforeseen spikes – bridal shops take note.
Sourced by Dirk Marichal, vice president EMEA & India at Avi Networks