AEMaaCS Cache Warming – Best Practices for Multi-Pod Dispatcher and Fastly CDN

Hi Everyone,

We are using AEM as a Cloud Service (AEMaaCS) with Adobe-managed Fastly CDN and 2 Publish/Dispatcher pods in our Dev/Stage environments.

We have implemented a cache warmer scheduler for around 150 dynamic pricing pages (non-JCR pages) that fetch data from Magento. These pages take around 3–4 seconds to load on a cache miss.

What we observed

• Each Dispatcher pod maintains its own local disk cache.

• Cache warming requests through our public domain are distributed across both pods.

• During subsequent runs, URLs cached on the same pod return quickly, while requests routed to the other pod result in cache misses and take 3–4 seconds.

• Running the warmer twice helps populate the cache across both pods.

Questions

1. Dispatcher Cache Warming: Is running the cache warmer multiple times the recommended approach to warm all Dispatcher pods, since we cannot target individual pods directly?

2. Fastly CDN: How can we warm the cache across multiple Fastly POPs (Sydney, Melbourne, etc.)? Does Fastly shielding in AEMaaCS help reduce cache misses across POPs?

3. Execution Approach: Is running Sling Jobs with options.onLeaderOnly(true) on Publish the recommended approach, or should we use Author or an external scheduler such as AWS Lambda?

4. Best Practices: Are there any recommended approaches or documented best practices for cache warming in AEMaaCS, especially for dynamic pricing pages?

Would appreciate any suggestions or real-world experiences from teams using AEMaaCS.

Thanks

Vishnu