| Item type |
SIG Technical Reports(1) |
| 公開日 |
2022-09-06 |
| タイトル |
|
|
タイトル |
eBPF-based Packet Tracing for Service Mesh |
| タイトル |
|
|
言語 |
en |
|
タイトル |
eBPF-based Packet Tracing for Service Mesh |
| 言語 |
|
|
言語 |
eng |
| 資源タイプ |
|
|
資源タイプ識別子 |
http://purl.org/coar/resource_type/c_18gh |
|
資源タイプ |
technical report |
| 著者所属 |
|
|
|
Toyota Motor Corporation |
| 著者所属 |
|
|
|
Hosei Uniersity |
| 著者所属 |
|
|
|
Hosei Uniersity |
| 著者所属(英) |
|
|
|
en |
|
|
Toyota Motor Corporation |
| 著者所属(英) |
|
|
|
en |
|
|
Hosei Uniersity |
| 著者所属(英) |
|
|
|
en |
|
|
Hosei Uniersity |
| 著者名 |
Chunghan, Lee
Reina, Yoshitani
Toshio, Hirotsu
|
| 著者名(英) |
Chunghan, Lee
Reina, Yoshitani
Toshio, Hirotsu
|
| 論文抄録 |
|
|
内容記述タイプ |
Other |
|
内容記述 |
Large scale microservice applications have been arisen and the application's architecture and container overlay networks on servers have also been complex. Although the distributed tracing for the service mesh is widely adopted in the microservice applications, it only focuses on latency-based monitoring and service discovery at an application layer. It is still challenging to monitor the container overlay networks to collaborate with the service mesh. In this paper, we present a packet tracing method using eBPF for the latency measurement based on the service mesh on the container overlay network. To detect the distributed tracing context on a HTTP header efficiently, we move the location of the tracing context at the head of the HTTP on sidecar proxy. Our tracing method gathers the HTTP packets that have the tracing context and measures the latency using eBPF. Our evaluation using an open-source benchmark on Kubernetes shows that there is no significant change of end-to-end latency using the proposed tracing header format. Moreover, our eBPF tracing method presents the similar latency characteristics on the container overlay network in comparison with tcpdump. |
| 論文抄録(英) |
|
|
内容記述タイプ |
Other |
|
内容記述 |
Large scale microservice applications have been arisen and the application's architecture and container overlay networks on servers have also been complex. Although the distributed tracing for the service mesh is widely adopted in the microservice applications, it only focuses on latency-based monitoring and service discovery at an application layer. It is still challenging to monitor the container overlay networks to collaborate with the service mesh. In this paper, we present a packet tracing method using eBPF for the latency measurement based on the service mesh on the container overlay network. To detect the distributed tracing context on a HTTP header efficiently, we move the location of the tracing context at the head of the HTTP on sidecar proxy. Our tracing method gathers the HTTP packets that have the tracing context and measures the latency using eBPF. Our evaluation using an open-source benchmark on Kubernetes shows that there is no significant change of end-to-end latency using the proposed tracing header format. Moreover, our eBPF tracing method presents the similar latency characteristics on the container overlay network in comparison with tcpdump. |
| 書誌レコードID |
|
|
収録物識別子タイプ |
NCID |
|
収録物識別子 |
AN10444176 |
| 書誌情報 |
研究報告システムソフトウェアとオペレーティング・システム(OS)
巻 2022-OS-157,
号 9,
p. 1-7,
発行日 2022-09-06
|
| ISSN |
|
|
収録物識別子タイプ |
ISSN |
|
収録物識別子 |
2188-8795 |
| Notice |
|
|
|
SIG Technical Reports are nonrefereed and hence may later appear in any journals, conferences, symposia, etc. |
| 出版者 |
|
|
言語 |
ja |
|
出版者 |
情報処理学会 |