An LTM Specialist upgrades the switching infrastructure and the backend servers on the LAN segments.
The LTM Specialist notices a 20% memory usage increase on the BIG-IP device while handling the same number of concurrent connections.
A comparison of statistics pre-upgrade and post-upgrade shows a significant reduction on the following:
-RTT between the BIG-IP device and the backend servers
-Packet drops in the switch
Time to First Byte (TTFB)
The LTM Specialist is concerned with the scalability of the number of concurrent connections with the new memory usage.
Which setting should be changed to reduce the memory usage on the BIG-IP device?
Recent Comments (The most recent comments are at the top.)
When the network quality improves (reduced RTT, fewer packet drops, and lower TTFB), backend servers send data faster to the BIG-IP device. If the client’s receiving speed is slower than the server's sending speed, the BIG-IP device temporarily stores this extra data in its memory (using proxy buffers). This is why memory usage increases.
By reducing the proxy buffer high setting, the BIG-IP device will store less data in memory before forwarding it to the client, thereby lowering memory usage.
Example:
Pre-upgrade: Servers send data at a moderate speed, and the BIG-IP device doesn't need to store much data in memory.
Post-upgrade: Servers send data much faster, and the BIG-IP device stores more data in memory because it can’t send it to the clients quickly enough.
Reducing the proxy buffer high setting tells the BIG-IP to forward smaller chunks of data to the clients instead of holding onto large chunks in memory, solving the memory usage issue.