Application Acceleration, Optimization, and Traffic Management
This domain covers compression, caching, front-end optimization, TCP and HTTP tuning, content switching, and overload controls. The exam tests when an optimization helps and when it breaks an application.
Core rules
- Compression reduces response size for text-based content such as HTML, JSON, and CSS. It is ineffective for images, video, and already-compressed files.
- Integrated caching stores responses according to cache policies. Respect Cache-Control and avoid caching personalized or authenticated content.
- Front-end optimization rewrites HTML, CSS, and JavaScript to reduce requests. It can break complex applications, so test before enabling broadly.
- TCP profiles tune window scaling, SACK, keepalive, and related settings. HTTP/2 provides multiplexing on the client side and can reduce latency.
- Content switching selects an LB virtual server based on host, URL, or other request data. A default virtual server handles unmatched traffic.
- Surge protection, priority queuing, and QoS manage overload and protect critical traffic.
Worked example
An API returns 200 KB JSON responses, and static assets are served from the same VIP. Enable compression for application/json above 1 KB. Do not compress JPEG images. Cache /assets for one hour, but do not cache /api because responses are personalized. Add a content-switching virtual server: /api goes to the API LB virtual server, /assets goes to the static LB virtual server, and a default virtual server returns maintenance content. The reasoning is that optimization must match content type and cacheability; applying compression or caching indiscriminately can waste CPU or leak user data.
Traps that cost marks
Compressing already-compressed content wastes resources. Caching personalized responses can expose one user's data to another. FEO can break JavaScript-heavy applications. Content-switching policy order matters, and a missing default virtual server produces errors for unmatched requests. A mismatched TCP profile can hurt throughput. HTTP/2 benefits depend on client support and the backend path. SSL compression is generally disabled and should not be relied on.