Chapter 8. Performance Networking and Traffic Optimization
So far we have focused on getting access and connectivity working with Cilium. Outside of the eBPF-based kube-proxy replacement we covered in Chapter 6, we have yet to discuss network performance.
Our goal is not just to help you run Cilium; it is to help you get as much value and performance out of it as possible. A network tool is often judged on throughput, latency, and jitter, and the same goes for Cilium.
In this chapter, we will review some of the Cilium features that enable you to optimize network performance. We will not cover them all; to avoid overwhelming the reader with too much information, we decided to leave out topics like Bottleneck Bandwidth and Round-trip propagation time (BBR), BIG TCP, and eBPF host routing.
Note
Excluding topics is a difficult choice. To briefly summarize their value:
-
BBR is a modern congestion control algorithm that improves throughput and reduces latency on long or high-bandwidth connections.
-
BIG TCP increases efficiency for large-scale data transfers by allowing larger packet sizes, reducing per-packet overhead.
-
eBPF host routing bypasses parts of the traditional networking stack to deliver faster packet forwarding directly within the kernel.
The Isovalent blog covers BBR and BIG TCP in numerous posts. eBPF host-routing is enabled by default if your kernel supports it, so it does not require further explanation here.
We will focus on a set of features that boost performance ...
Become an O’Reilly member and get unlimited access to this title plus top books and audiobooks from O’Reilly and nearly 200 top publishers, thousands of courses curated by job role, 150+ live events each month,
and much more.
Read now
Unlock full access