It’s never too early to start thinking about the holiday shopping season, especially since most Americans are already anticipating outages and system failures from their favorite online retailers. According to a survey conducted by ScaleArc, a database software company, 52% of shoppers expect to experience an outage on days like Black Friday and Cyber Monday. To prepare for the busiest online shopping season, companies need to ensure their systems are ready for extreme scalability and continuous IT operations, year-round.
Every year more Americans are turning to online shopping, according to news reports. The shift to buying online has impacted brick-and-mortar sellers significantly. Big retailers including Gymboree and Lane Bryant are closing thousands of stores, and 25% of shopping malls are expected to close in the next 5 years.
Given increased reliance on – and expectations for – online shopping, it’s more important than ever for online retailers to prepare for the onslaught of traffic during holidays like Black Friday and Cyber Monday.
Yet despite online retailers’ major investments in technology, most Americans expect big retailers to have downtime or website crashes around the holiday season. According to ScaleArc’s survey of more than 1,000 Americans, respondents anticipate major retailers like Walmart, BestBuy, Macy’s and Target to have the highest risk of crashing.
While consumers take the last of their summer vacations, retailers need to make final changes to their infrastructure and test their systems to handle the surge of holiday shoppers at the end of the year.
It’s becoming even more critical that retailers prepare all year long for the holiday traffic surge. Several big-name companies botched Black Friday and Cyber Monday last year, including companies like Macy’s, QVC, Walmart, Victoria’s Secret, and Newegg. As we saw with these companies, even the smallest misstep in these important preparations can lead to major problems down the road.
When web and mobile sites buckle under huge volumes of traffic and experience outages, retailers lose revenue, and their brands suffer significantly. In the survey, 26% of online shoppers said they would leave a website if they experienced an issue, and 21% said they would visit a competitor's website. Customers don’t have time to wait around for a page to load —they’re already thinking about the next sale and how fast they can fill their carts and check out.
Preparing for the holiday season requires a continuous process. ScaleArc has developed a chronological checklist for retailers, designed to help IT and Ops teams target all 12 months and make sure the weakest links in the technology chain are eliminated.
Here’s what to do this month and in the months leading up to the holiday season:
August/September: During the August and September months, it is time to stop making changes to the infrastructure and start diving deep into analytics to see where potential weaknesses are in the system. Most companies should enter a freeze at this time and begin testing at high load, since it gives a company time to address any issues uncovered.
October: Companies should see traffic increase daily, which is good. Natural traffic growth can help with testing, and it provides a better test backdrop than load testing, typically because it reveals surprises. The team’s Black Friday preparations should be finalized and in place.
November: This month is the “Super Bowl” of e-commerce — where retailers can generate more than 30% of their yearly sales. Organizations with high Black Friday loads report Black Friday traffic being a 10X to 15X spike over rest-of-year numbers. Some companies will see as much as a 5X spike on Black Friday compared to Thanksgiving Thursday.
In November, the company can make any last-minute adjustments, but if you’ve been leveraging analytics throughout the year, they should be minor. Any final tweaks should be based on an unexpected traffic increase in October that signals higher traffic rates on Black Friday, according to the checklist.
December: It’s time to celebrate, but not for long. Evaluate what happened and prepare a postmortem. Now is the time to assess what went wrong, what broke, and what didn’t perform as expected. Robust analytics software with troubleshooting tools can help, along with a solution that provides a cluster-wide view of database server performance. Finding a permanent solution to any emergency fixes implemented over Black Friday should help you prepare for the next holiday season.
The information you gather in the postmortem is important; you’ll also need it to make a schedule for January priorities.
The database tier can be an especially problematic area to explore. Detailed analytics help reveal which application servers generate the most database load and what type of queries they execute, right down to query fingerprints with precise performance data, so you can focus your attention on items that will yield the biggest improvements.
December is also the time to evaluate and secure a budget for the upcoming year. An infrastructure investment will pay for itself and much more, and while the income numbers are still at the top of everyone’s mind, plan that budget.
Plus, if you look at quarterly revenue, a large investment doesn’t make sense, but if you look at yearly figures, that’s where you’ll see the ROI – the numbers during Black Friday can help justify a later expense.
January/February: Use these months to baseline your infrastructure and build for scalability. Define success criteria for the organization, understand how all parts of the infrastructure perform and identify the weakest links. You can’t improve what you can’t measure, so collect data on system performance. For example, how did the app servers hold up, what happened to the database clusters, and how well did the web servers perform?
According to many IT infrastructure execs, the database still remains one of the hardest parts of the infrastructure to scale. Even though databases may have enabled scale out capabilities, it’s the other parts of the infrastructure (like the application tier) that remain a challenge to scale out.
Even if you suffered no major failures, you need to understand how all parts of the infrastructure performed and build an infrastructure that’s ready to scale for traffic and transactions that go beyond your expectations.
After the analytics are complete, you can make a road map for the next season.
March/April: While the infrastructure may perform well during these months, it doesn’t mean it can handle the Black Friday load in November. Retailers need to ensure that the infrastructure can scale beyond expectations of traffic and incorporate a high-availability strategy that delivers resiliency in case of server failure, according to the checklist.
Web load balancers have proven invaluable for enabling resiliency at the web tier. With connection management, load distribution, and the ability to detect and route around downed web servers, these devices have enabled zero downtime at the web tier.
During these months, investigate how analogous technologies can act as an abstraction layer between your apps and your databases, shielding your customers from database downtime and enabling zero downtime at the application tier.
May/June: During the May and June months, consider upgrading the infrastructure or architecture. Even if certain technologies or features are unrelated to the Black Friday efforts, they still should be in play just in case they affect the entire system.
There’s a 24/7/365 spotlight on e-commerce IT infrastructure, which means seamless and immediate database failover is central to achieving zero downtime. Keeping your customers logged in and productive is paramount to Black Friday sales numbers.
A company should focus on technology that not only helps avoid downtime during unplanned failovers but also eliminates downtime from maintenance windows. Aim for zero downtime patching on database servers, with software routing around the offline database servers.
July: This is a busy month, and it’s all about supercharging performance so it can handle the Black Friday traffic loads. Databases typically fail because of poor query performance, lack of concurrent capacity to handle high user loads, and connection pooling and management issues. A company’s technology should address those challenges with a few capabilities, like query caching, the ability to add database nodes with no application code changes, and connection multiplexing.
Caching built into database load balancing software means you can apply the benefits of caching with no application code changes. Companies also need to ensure the SQL query cache is fully ACID compliant.
Retailers should also look for help with connection pooling and multiplexing that doesn’t require application changes. Reducing the number of connections to the server and keeping connections persistent can boost database access performance even when other speed enhancements, like caching, aren’t enabled.
Another capability that can address challenges during this month is adding additional nodes to your database infrastructure to increase system performance or disk space. You want applications to be able to leverage additional database servers as they come online – look for software that helps load balance the database traffic across all available nodes and keeps your eCommerce applications humming and your customers online.
Again, now is the time to move testing into high gear but use every month of the year to prepare for the holiday season to keep your systems running and avoid frustrated shoppers and loss of revenue.