Three Practices for Keeping Your Site Performance Fast and Stable
With online traffic surging, Arc XP offers three key practices for maintaining site performance to meet growing demand.
By Jeana Garms | 09/24/2020
Earlier this year, as consumers and organizations did an about-face to conduct work and life from the confines of home, digital activity experienced an unprecedented surge, with traffic estimated at 50 to 70 percent higher than previous levels.
Such demand has focused a spotlight on the need to re-evaluate website infrastructure to ensure not only site stability, but accessibility and speed.
Amid this record-setting traffic, Arc XP’s digital experience platform has kept client sites running fast and reliably, supporting more than one billion monthly unique visitors around the world and delivering eight billion page views per month.
To support our clients’ sites, many of whom have seen record increases two to three times their normal traffic volumes, we follow three overarching best practices that Arc XP was built, and continues to operate on:
- Design for Performance
- Monitor Your Applications and Infrastructure Regularly
- Focus on A Culture of Collaboration
Design for Performance
The first step to performance and stability is ensuring your site is efficiently built. Keep page weight light to improve page load speed and reduce bandwidth consumption while delivering a better user experience. For developers, this means understanding the importance of site performance alongside the functionality they are tasked with building.
“So many design decisions can influence a site’s performance,” says David Zayas, Vice President of Platform Architecture for Arc XP. “Developers need to ask themselves: ‘How well am I optimizing assets? How cacheable is my content? How is my HTML built? What libraries am I using, and when?’ These may seem like small considerations, but across a large site, they can end up making a big difference.”
For every second a page takes to load, there are real dollars associated with whether a user stays on the site or not. The more frequently your system has to transfer large data bundles – answering requests for images, styles and java files for instance – the greater the impact on site response time.
In addition to optimizing assets, keep pages light by using techniques such as lazy loading to reduce the number of external site calls and employ a content delivery network (CDN). Common to an extent, surprisingly less than 75 percent of the top 100,000 sites are utilizing CDNs today, highlighting the opportunity to leverage CDN support.
If you’re not currently using a CDN, you’re missing out on caching repeatable actions and pages that rarely change, leaving your site to handle every request through its infrastructure. As traffic levels increase, so can system stress, slowing page load speeds down to subsequently deliver a poor user experience.
Designing for performance also means not creating a monolithic system that can only be tuned as a whole. To this end, Arc XP is built on a series of microservices, each with its own infrastructure and its own set of systems and monitors. Through this configuration, when a fix or adjustment is needed, we’re able to address that specific application. “We regularly see this as a missed opportunity. Every application is slightly different, so it’s important to have the ability to tune and scale them individually rather than adjusting the whole system,” says Zayas.
Monitor Your Applications and Infrastructure Regularly
Equally as important as designing a lightweight site is deploying monitoring systems that alert your operations team when something goes wrong with your infrastructure – or better yet, warns you before a problem occurs.
Set up dashboards that interpret network data to reveal traffic trends, so that when spikes occur, its location can be easily identified and resources can be scaled up, horizontally or vertically as needed.
Using monitoring insights to regularly run test scenarios will provide some aspect of predictability for where you’ll need to scale, and if you don’t have the automation in place, where you may need human resources to scale manually, increasing the importance of early warning tools for avoiding performance issues before they happen.
For Arc XP, our engineers are able to monitor and measure platform behavior at its deepest levels. “We’ve built a series of tools that collect metrics within Amazon Web Services, using services like CloudWatch and CloudTrail. We bubble all that data together so we can aggregate everything that’s happening across the platform.”
Using a collection of dashboards and a tool called Datadog, we monitor how long each fetch takes and counsel clients on how they might improve performance. “We’ve set up comprehensive monitoring systems that we’re constantly watching,” says Zayas. “If we notice that a customer’s page is taking too long to load, or is coming back with an error state, we get all that data ahead of time. Then we triage, figure out the cause and get ahead of it proactively.”
The monitoring dashboards and alerts are just one side of the site performance coin. The other aspect is user experience. Tools such as Real User Monitoring and synthetic testing deliver the performance insights needed to ensure your site is delivering the expected user experience.
Approaching user experience monitoring with the same level of performance rigor and visibility that we’ve applied to our platform infrastructure, we partnered with digital experience monitoring provider Catchpoint to setup key user flows. Continuously tested and monitored, these user flows enable us to replicate the user experience and identify any corrective actions needed to meet experience and performance expectations.
Often forgotten is that site traffic is not limited to serving pages externally. For major news sites like The Washington Post, where Arc XP originated, internal audiences can generate significant traffic, especially in major news moments. When news breaks fast and frequently, a large team of content creators and editors can tax an infrastructure from the inside.
Managing the impact of internal traffic requires the same level of monitoring and techniques used for external site traffic. Leverage the frameworks, standards and monitoring tools in place to proactively access scalability needs to ensure workflow performance, especially in times of increased internal content creation and delivery frequency.
Focus on A Culture of Collaboration
The last leg of Arc XP’s three-legged site performance stool is perhaps the most important – culture.
Performance starts with culture. Too often business, product and engineering teams are siloed, operating their own sets of priorities and performance metrics.
It’s not uncommon for business stakeholders to approach tool selection or site design without consideration for the IT aspects, while IT has separate decision makers who are not familiar with the business or consumer needs. These silos result in waste, inconsistent production and missed opportunities.
Breaking down barriers to forge a culture of cross-functional communications and relationships is paramount to driving site performance. This collaboration needs to occur not only at the strategy level but carry through execution. Having the business and product working in lockstep with engineering will result in common site improvement goals that drive greater engagement, reduced bounce rates and overall increased conversions.
For our part, whether we’re approached by the business stakeholders or are engaged by the engineering team, we always ask to have the other key constituents brought to the table. Through this we have found that many times, the process is able to move quicker with a much higher rate of success.
Entering A New Age of Digital Reliance
While digital activity is likely to come down, it certainly will not return to pre-Covid levels as businesses create lasting remote work policies and ecommerce continues to accelerate with consumers changing shopping habits.
This means that while site performance has always been important, now it’s critical to business success. Modeling your site and infrastructure to follow the key practices outlined above will well position your organization to deliver a stable site that meets user’s performance expectations.