Every year, Black Friday is used as a barometer for the overall health of retail and e-commerce. One of the interesting insights that came out of this year’s Black Friday came courtesy of Web performance company Catchpoint System’s report that e-commerce websites had sharply degraded in performance between 2013 and 2014, with desktop-centric e-commerce pages stumbling to a 20% performance hit while mobile pages slowed by 57%.
Not exactly the best way to set the cash register on fire. Not in a good way, anyway.
The problem, it turns out, is scale. Or, rather, online retailers’ ability to scale. To get some insight into modern scale challenges for retail as we close out the holiday season, I spoke with Kathryn Murphy, senior vice president of Apps and Platform at Tomax, a Salt Lake City retail software company that specializes in brick and mortar retailers (i.e., L.L. Bean, Swarovski and Sportsman’s Warehouse), but also provides a suite of cloud-based retail solutions that support more than 25,000 customer store locations.
See also: As Systems Get More Complex, Programming Is Getting “Reactive”
Tomax is on the front lines with retailers, and has seen an increasingly “r” approach to programming pay dividends as they attempt to scale.
ReadWrite: What are the new pressures that you see in retail? What could explain the struggle of retailers to keep up with the demands of scale?
Murphy: The biggest pressure in retail is the shift in power to the consumer. Several years ago, consumers got pretty comfortable shopping online and on their mobile devices with online retail giants like Amazon. This quickly became the new expectation from the consumer.
Consumers don’t care how hard it may be to deliver that experience. They expect a big box retailer like Best Buy to have connected, real-time systems. The consumer expects to be able to order online, see the inventory in a local store, and pick it up later today. They expect the associates to have mobile devices and access to all the same information. All of these new devices, new endpoints, new interactions from the consumer and the associates have made the retail systems world unpredictable.
RW: How much of the year do retail IT pros spend worrying about and preparing for the holiday surges?
KM: For the highly seasonal retailers, it’s everything. As evidence, they all implement “lock down” periods in the fourth quarter. Everything must be done and ready for the holiday season by September.
Additionally, they all purchase infrastructure based on “peak season”—this is a phenomenon unto itself. So many retailers are dealing with orders of magnitude differences in system load during the holiday. Honestly, it’s a real opportunity for elastic systems that can scale when the retailer needs it – and pay only for use—instead of making enormous upfront capital expenditures and paying for peak load infrastructure all year just to handle the holiday season traffic.
RW: How has the proliferation of devices—in the hands of customers as well as employees—affected the retail landscape?
KM: It has two major effects. The employee devices have created hundreds of more endpoints in a retail environment. No longer is it a fixed number of registers and a few scanning devices in the backroom. Everyone has a device and they are doing all sorts of things from helping customers, ringing up transactions, and performing store inventory.
And these devices are not hard-wired into the environment, they are constantly losing connectivity, being replaced, etc. This has the effect of creating unpredictability and instability.
Then add the consumer and their new empowerment to perform transactions themselves, now we have that many more devices, and many different types of devices in the mix. And since it is the consumer’s device, they are free to interact with the retailer system at any time. This element alone puts pressures on typical retailer IT practices like “nightly system maintenance” windows.
RW: Why is resilience so important in retail?
KM: All of this technology is highly dependent on networks and system availability. Both of which fail.
And yet, a single failure for either a consumer or associate can be so impactful that it can result in a loss of a customer for the retailer. It’s critical to have “smart apps” that can gracefully recover and transition when failures occur without the user noticing or having to do anything special.
RW: In a recent video Tomax talked a lot about concurrency being a major challenge for retailers. What does concurrency mean in your retail stores? How does reactive programming help?
KM: In retail, we see some unique challenges with concurrency with the proliferation of devices. For example, when a big shipment comes in (and in the holiday season, these can be really big!), it seems obvious that you would want as many employees as possible scanning the merchandise in and preparing it for the retail floor.
Concurrency is immediately a problem since more than one employee is scanning merchandise, and sometimes even the same item, at the same time. The system needs to inform both employees, in real time, about the other person’s updates.
Traditional systems want to “lock” a record while one person works on it and make the other person wait. Reactive systems take a different approach and leverage work queues and threads to keep communication flowing freely back and forth.
We’ve had great success using Typesafe‘s reactive stack, as it frees retailers from having to train their users to use their systems in a certain way. Instead, this reactive approach makes it easier to build intuitive applications that respond to the technical environment around them, rather than forcing users into the system’s rigid way of doing things.
Lead image courtesy of Shutterstock
