What happens to Hybris at peak periods?

August 19, 2019

Hybris is extremely fast and scalable, but what happens to Hybris at peak periods? Ideally, you want your B2C eCommerce site’s uptime to exceed expectations. However, like any software system, Hybris is restricted by database limitations, and transaction-heavy pages, such as checkout, can be overwhelmed by traffic at peak periods.

Whether customers are shopping for Christmas, Mother’s Day or mid-season sales, eCommerce sites can experience high pressure at the checkout stage, sometimes causing purchases to grind to a halt mid-transaction. The impact of increased traffic – delays, or even downtime – can be a massive issue in any eCommerce scenario.


Clearly, technical problems during key times such as Black Friday and Cyber Monday can leave customers with a poor impression of your business if your site reaches a painfully slow speed or even goes down altogether. The good news is that, as a scalable, cloud-hosted platform, Hybris can handle increased traffic during peak periods – provided that you’ve prepared to make sure that this is sustainable.


In readiness for key retail dates, eCommerce sites need to be ‘peak-proof’ well before time – that means allowing months rather than a few weeks for capacity planning and making testing a continuous process. If you’re using Hybris to power your eCommerce site as a full end-to-end solution, it’s critical to optimise it for an exceptional user experience and ensure that you have enough resources. Speed and scalability are the most important KPIs – these impact not only conversion rates, but your site’s SEO – a sometimes-overlooked by-product.


Here’s what you should look out for and address when capacity planning to keep your eCommerce store stable, even at peak times when the volume of online consumer traffic is at its highest...


High page load times

Page speed – the time it takes to fully display the content on a particular page – can be an issue during peak periods. It makes for a bad UX as there’s little more frustrating than waiting for what feels like aeons for a page to appear. Unsurprisingly, slower page time leads to increased page abandonment.


How many seconds should it take for a page to load, ideally? You need to aim for a one-second page loading time. Page speed impacts on site speed, which is the amount of time taken for a sample of page views on a site. Google now pays particular attention to site speed on mobile as well as desktop, and a slow site is, in fact, a negative ranking factor for SEO. None of this should be an issue, though, if you’ve fine-tuned your system and made sure that you have enough resource.


Current and average disk queue length

You can pinpoint possible bottlenecks that are likely to occur during peak periods by monitoring the number of requests represented by current and average disk queue length, as well as paying attention to the Pages Per Second (PPS) counter. PPS is the rate at which pages are read from or written to disk to resolve hard page faults.


This process will help you to assess if there’s a risk of the memory system being negatively affected. If the rate at which your system is paging memory to or from the disk is too high, it can result in faults that can cause system-wide delays. Memory is often the cause of performance issues, so it’s a good place to start when you’re troubleshooting. You can read more about that here.


Monitor and test

To make sure that your site is reliable and scalable and look after your revenue and reputation by delivering the best solution to your customers, you should aim to test and fix continuously. It’s critical to keep a close eye on how it’s working day to day, so that you have a consistent, shared view of performance. Using free tools such as SAP Customer Application Lifecycle Framework (ALF) for Commerce to develop a best-practice approach is recommended.


Create a dedicated, isolated testing environment and be sure that you test the full stack, from infrastructure components to applications, including third-party integrations (for example, payment provider, shipping provider, tax provider). Third-party analytics tools such as TradeState and SharePath can also help you to achieve end-to-end monitoring, helping to identify where things are slowing down so you can take action.


Elastic performance management

Consider your hosting arrangement and assess whether you’d benefit from the increased flexibility of elastic cloud computing technologies that allow your systems to scale based on demand. If you’re regularly reacting to peak periods, the ability to up your computer power as needed can be a great advantage.


Moving from on-premise hosting to the cloud removes complexity and increases agility and hosting solutions are available across all cloud services, including: Amazon Web Services (AWS), Microsoft Azure and Google Cloud Platform (GCP).


As these platforms vary in capability, you may wish to enlist external help to understand how to get the most out of a scalable and resilient platform that allows you to pay for computer power as required.


Aim for the best, plan for the worst

Having a contingency plan ready will ensure that you’ve put in the necessary time and thought as to what could go wrong. This process of risk assessment will keep your team on its toes and highlight any potential weak spots – make sure that you test these well in advance! Look at your catalogue design, customisation, integrations and configurations for a deep dive into what could go wrong.