Prepare your site for traffic surges, all year long

The 2020 holiday season goes down in history as the busiest ever for ecommerce. Although ecommerce sales have been on the rise for years, the pandemic has created a surge in online shopping. First up, there was Black Friday, where U.S. consumers spent a whopping $9 billion online. This shattered previous Black Friday records, with sales up 22% from 2019. Then came Cyber Monday, where consumers went a step further and spent $10.8 billion in one day, marking the biggest online shopping day in U.S. history.

Too much traffic to handle

Unfortunately, the sheer volume of traffic was too much for many websites. Even some major brands experienced performance issues on their websites on the Black Friday weekend, with many more happening on Cyber Monday.

Website performance issues have a significant impact on revenue. Online audiences have a low tolerance for poor website performance. Even just a second or two of loading time can lead consumers to immediately go elsewhere. According to the latest State of Online Retail Performance report, a two-second page-load delay can increase bounce rates by 103%. And those audiences are less likely to come back to your site again, with 79% of shoppers being less likely to buy again from websites with poor website performance.

Preparing for high demand in 2021

It’s hard to predict what will happen in 2021, but there are no signs of ecommerce slowing down anytime soon. Even if your resources are lean, there are some budget-friendly and relatively straightforward things you can do to prepare for high traffic. Here’s what we recommend:

Step 1: Perform load testing

2-4 months before your peak-traffic period begins, depending on environment size and complexity

• Determine your expected peak traffic, based on your analytics and marketing data. It’s reasonable to expect a 20-30% increase in traffic year over year.
• Test for at least 125% of your expected peak traffic.
• Be thorough and cover all functions of your application, including integration points.
• Aim to test against an environment that is as close to production as possible with the same codebase. In modern, complex applications, load does not scale linearly — and real-world performance cannot always be extrapolated out accurately.
• Look for a load testing tool that incorporates realistic “think time,” so you can more accurately gauge the impact of real users.
• Establish your goal metrics (such as 95th percentile for transaction completion).
• Allow time to scale your infrastructure, if needed.
• Conduct multiple load tests, to confirm your results and ensure you’re able to achieve stability at peak user traffic.

Step 2: Review your integration points

1-2 months before peak traffic period

• Review both internal and external integrations, such as search, payment gateways, OMS, inventory, mobile, etc.
• Review your architecture, where possible.
• Make sure integrations are highly available.
• Determine the end-user impact if the service were to become degraded or unavailable.

Step 3: Review your configurations

4-6 weeks before peak traffic period

• Proactively review your configuration and application performance management (APM) metrics.
• Make sure to review not just the application itself, but also web servers, caching layers, databases, etc.
• Allow time to make any necessary changes and test if necessary.

Step 4: Conduct a configuration review withing your APM and monitoring tools

4-6 weeks before peak traffic period

• Review any dynamically calculated baselines.
• Review business transaction configurations in APM tools.
• Ensure health rules are appropriately configured for the environment.
• Review dashboards to reduce troubleshooting time in the event of issues.

Step 5: Review your network layer

2-4 weeks before peak traffic period

• Review performance on firewalls and load balancers and any inline devices (such as IDS, etc.), to ensure they’re able to handle increased traffic.

Step 5: Tune your CDN/Caching

2-4 weeks before peak traffic period

• Be sure as many static assets as possible are cached.
• Ideally, the CDN/cache layer should take as much of the traffic as possible, reducing calls back to the application.

Step 6: Review any scheduled jobs to minimize impact during peak hours

1-3 weeks before peak traffic period

• Scheduled jobs might include cronjobs, scheduled tasks, backups, etc.

Step 7: Review your URL monitoring

1-3 weeks before peak traffic period

• Synthetic transactions should be in place to monitor availability and functionality of critical business paths (e.g., add to cart, checkout, search, etc.)

Step 8: Review your guidelines

1-2 weeks before peak traffic period

• Verify that escalation paths, return-to-service instructions and runbooks are up to date and technically valid.

Step 9: Prepare your team

1-2 weeks before peak traffic period

• Limit code deployments leading up to the peak traffic period to minimize changes in the environment.
• Consider limiting or disabling backend admin functionality during peak traffic periods, such as imports, catalog updates, etc.
• Consider scheduling regular, short touchpoints throughout the peak period with all appropriate parties.

Get help when you need it

The experts at Rackspace Technology are here to help get your applications running at optimum performance. We’re here to provide strategic guidance, planning and implementation services, plus day-to-day troubleshooting support — so you can drive improved performance and reliability, all year long. Reach out to our digital experience experts to get started.