Global online sales are projected to reach a record high during the 2015 holiday shopping season, growing 11 percent from 2014 to reach $83 billion in total sales this year.
As the frequency of online purchases and higher value transactions continues to be a growing trend amongst online shoppers, it is clear that during the holidays, online shoppers will be more willing than ever to make bigger purchases.
This news shouldn't scare e-tailers, but the reality is that many still live in mortal fear of the kind of site problems that often emerge during this busy season-slowdowns, checkout errors, crashes-which can derail holiday hopes. While today's ecommerce companies understand that the holiday season will stress their websites and that they should closely monitor and prepare their sites for these eventualities, they don't precisely understand how to do so.
The stakes are even higher on days such as Black Friday, since many ecommerce companies choose to mark down their entire inventory to include in the sale, as opposed to only a few select items. E-tailers who fail on Black Friday risk losing 10 to 15 percent of their annual revenue. While that big sales day has passed, the following tips can ensure success for ecommerce merchants during the reset of this crucial sales period.
Ecommerce companies that rely on a legacy relational database like MySQL have an especially difficult time accommodating holiday traffic. This is due to the fact that in order to handle the increase in load, companies have to purchase costly additional hardware to "scale up." When they can't accurately gauge the level of resources they will actually end up needing, they end up over-provisioning. If companies under-purchase, they risk the deluge of shopper purchase activity crashing their database. If they over-buy hardware for high-stakes days such as Black Friday or Cyber Monday, they will be stuck with superfluous resources that ultimately amount to a sunk cost.
Companies can mitigate this risk by powering their site with a scale-out relational database that provides the ability to increase capacity by simply adding commodity servers. This enables you to incrementally add or remove capacity to and from your database, so that you can much more cost- and resource-effectively adjust to the demands that will be placed on your site over the holidays, or other peak periods. Scale-out relational database that can run in the cloud, makes it even easier to scale as needed.
In addition to the massive spikes in traffic during the holiday shopping season, ecommerce merchants must also consider the influx of first-time visitors to the site, who will exhibit new behaviors and interact with the site in different and unfamiliar ways than the regular visitors do. Special sales such as flash-sales or daily deal sales can further skew the behavior of customers in new and unexpected ways.
By running a pre-sale ahead of upcoming important days such as Green Monday, Free Shipping Day and Super Saturday, merchants can model customer's behavior without reaching Green Monday or Super Saturday levels. This way, they are able to monitor these new behaviors in lower volumes and examine how the site reacts, so that they are able to make the necessary preparations and adjustments ahead of the big day.
A variety of different factors can cause a database to overload, and it is important to prepare for the ways in which they might crash in order to know which route you should take to solve the problem and get it back up and running as quickly as possible, especially during times of peak traffic such as the holidays.
But what happens if you crash?
It is important for ecommerce merchants to be able to quickly recover in the event of an outage. A Disaster Recovery (DR) database is important to getting ready for the holiday shopping rush. Should the database crash, you'll likely need to scale the DR database to be able to handle a higher load than the master. After all, if the production database was overwhelmed by traffic or checkouts, switching to a DR of exactly the same size will simply result in another crash.
A DR is costly, but a failing site is more expensive than anything else, as it results in an extreme loss of revenue and potentially, a hit to the merchant's reputation. Ahead of the holidays, e-tailers must ask themselves, "How much is the security of knowing my website is less likely to fail during Super Saturday worth?"
While most people don't favor having to work during the holidays, it is important for ecommerce companies to have staff on call who are familiar with the infrastructure and the designated course of action to take should a problem occur with the merchant's ecommerce site including the infrastructure.
Even for a well-tested system, it is crucial to have people to track everything that is happening with the database in real time, so that the issues can be resolved and the database restored in a way that fixes the problem. For example, should it be noticed that the database is running out of resources due to the influx of holiday traffic, proactive measures can be taken to spin up more servers with greater CPU or memory capacities.
Ecommerce sites contain a myriad of features aimed at improving the user experience. However, these features can have the opposite effect during times of peak traffic. For example, a website that sells to both B2B and B2C customers might have a plug-in that helps determine whether a visitor to the site is a potential B2B or B2C customer to display targeted products. But this tool may add 10 percent of the website's page load time or resources to function, and may not be necessary largely consumer holidays such as Super Saturday or Free Shipping Day.
If for Super Saturday the company can anticipate that they are only offering discounted items to their B2C customers, they may consider temporarily disabling features, such as the aforementioned plug-in, to free up resources and improve site performance during the influx of traffic. Similarly, for holiday specials and eblasts, a static landing page gives a low resource / high performance entry point for customers. It allows them to get onto your site and begin browsing no matter your load, while also skipping more resource intensive dynamic browsing. This means more customers getting directly to the products they want with less server load.
We may never be able to predict or test for every scenario, but these tips should help ensure that companies can be safe rather than sorry come the holidays.
Mike Azevedo is the president and CEO of Clustrix where he brings more than 25 years of sales and executive leadership experience in scale-out analytic applications, grid computing, storage infrastructure, security and retail.