ec4d11e3e4b7be8f2e74981f70cc859f_.jpg

coscon23kv.png

菁蓉汇

2023 第八届中国开源年会

compress.jpg

With the flourishing development of cloud-native technologies, an increasing number of enterprises are opting to migrate their applications to Kubernetes. Kubernetes brings convenience for developers and operators, but it also makes troubleshooting more complex. When application failures occur, the root causes can stem from within the application itself, the infrastructure layer, or even the operating system kernel. Although traditional observability systems have solutions for application, container, and infrastructure layers, the data tends to be fragmented, lacking effective correlation.

The eBPF technology emerges as a powerful kernel technique that allows collecting runtime data from kernel hook points without the need for modifying kernel code or adding kernel modules. This speech will introduce how the eBPF technology enables full-stack observability of applications, containers, and the kernel, as well as the correlation and fusion of various observability data. Currently, there are numerous eBPF-based observability system implementations in the observability field, but most lack tracing capabilities. This speech will also cover the implementation of eBPF tracing and Alibaba Cloud's practices in this domain.

PresentationeBPF在云原生可观测领域的探索与实践