#97 Accelerating Kubernetes with In-network Caching


More

  • None

Accepted

[PDF] Final version (564kB) Sep 26, 2022, 10:55:43 AM UTC · 7873d8669b4fdbcb7d41897d4c8c2063cc70e2325ffd4cb3b8651b063b0e164e7873d866

[PDF] Submission version

We present a new Kubernetes architecture that leverages in-network caching to accelerate one of Kubernetes' core components, its key-value store. We also identify performance limitations of previous in-network caching platforms and propose a new platform that demonstrates better throughput and scalability by utilising a different replication method.

S. Sagkriotis, D. Pezaros
Eligible for SRC? (lead author is a student)

If eligible for SRC, specify student type

Graduate

Submission Type

Poster

  • Data-center
  • Performance
  • Programmability
  • Software/virtualization

ACM keywords:

Data plane; Replication; Kubernetes; P4

ACM Computing Classification

To edit this submission, sign in using your email and password.