Introduction
Promotions are a strategic tool for customer engagement, driving higher transaction volumes, and fostering customer loyalty.
However, they also present a significant risk of Denial of Service (DoS) outages if proper system preparations are not implemented to handle the abrupt surge in traffic.
This guide is intended for customers planning a promotion aimed at increasing web shop traffic. It outlines the necessary measures to ensure system stability and provide a reliable shopping experience.
This document also applies when executing performance tests.
References
Measures
Notify Intershop of Upcoming Traffic Peaks
To notify Intershop of upcoming traffic peaks, communicate the exact campaign days to your Customer Success Manager (CSM) or via Service Desk as soon as they are known. Provide estimated traffic and order counts per day and per hour, using historical data and analytics to forecast expected demand, while also considering trends and external factors that may affect demand.
Notifications should be made at least 7 days in advance to allow for operational preparations. Include essential details such as the system, channel, date, time, and duration of the campaign. Submit additional details (forecast) in an XLS or PDF document.
Review Page Cache Configuration
During high season, the number of concurrent website visitors and transactions increases dramatically. A page cache configuration that is prepared for this peak load can ensure that your site remains stable and fast, even when traffic surges. Therefore, review the page cache configuration for campaign entry pages before the high season begins. This is a task for the development team. Page caching should be implemented for all frequently accessed storefront browsing pipelines. In addition, any campaign-related URL parameters that are not relevant to the page content should be added to the page cache ignore list (webadapter.properties -> pagecache.ignore.<no>).
Also, be sure to avoid invalidating the full page cache via the back office during high-user hours.
Deployment of Code Changes / Releases
Do not schedule release changes close to campaign launch and during high season.
Any deployment that involves code changes runs the risk of causing side effects and negatively impacting platform functionality. Since this limitation has a direct impact on buying behavior and conversion rate, version changes should not be made during and just before the launch of the high season. Code adjustments should only be made for critical issues and in exceptional cases in the form of hotfix deployments.
Execution of Replication Processes
If possible, run replication processes only during off-peak hours. Replications cause additional system load and response time spikes because the entire page cache is cleared and the application server caches are refreshed at the end of each replication process.
Execution of Import and Search Index Processes
As with replication processes, import processes and search index maintenance processes should only be run during off-peak hours. Import processes cause additional system load and response time spikes because caches are also invalidated, which can increase application server load.
Catalog Filter Publishing
The publication of catalog filters should be configured to take place at times of low traffic so that the load on the application server and database is not affected at times of high traffic.
Robot Detection
Robots can cause unwanted additional load on a webshop and even block the system.
Before high traffic periods you should make sure that ICM can detect enough potential robots via a web server robots.txt file.