Leveraging Cloud Native KV to Build Hyperscale Infrastructure

Cloud Native @Scale
Tue, Aug 30, 2022, 10:00 AM (PDT)

About this event

Building large clusters (> 10,000 nodes) supporting web-scale applications can be challenging, considering the amount of metadata (e.g., container and node states) that need to be stored. However, it can still be possible with the help of the modern distributed KV database.

KubeBrain is a horizontally scalable coordination service for Kubernetes metadata storage, which abstracts the metadata storage interface required by the APIServer and uses a distributed KV database as the storage backend. Compared to Etcd, KubeBrain increases both read (~ 3 times) /write (~ 3 times) performance and can support clusters up to 50,000 nodes.

TiKV is an Open Source distributed transactional KV database providing horizontal scalability, high availability, and strong consistency, which makes it an ideal database for the next-generation cloud-native infrastructure. As a CNCF graduated project, TiKV has been widely adopted by many data management systems.

In this meetup, we will present challenges of scaling up Kubernetes to more than 5,000 nodes as well as how we can combine KubeBrain and TiKV to address them.

📅 Date: Tuesday, Aug 30, 2022

⏰ Time: 10:00 AM PST

📍 Venue: Zoom Webinar


🔶 Charles Zheng, Software Engineer at PingCAP

Chao is a Software Engineer at PingCAP with ~5 years of experience building large-scale cloud native systems. Chao is also a maintainer and member of several Open Source communities, like Kubernetes, OpenYurt, and KubeWharf. Chao has been a regular attendee and speaker at many academic conferences (i.e., SoCC, ICDCS, etc.) and industry forums(e.g., KubeConf).

🔶 Yixiang Chen, Software Engineer at TikTok

Yixiang is a Software Engineer at TikTok, building a cloud-native platform supporting applications serving TikTok ~10 billion customers worldwide. Yixiang has been one of the core architects of TikTok's cloud-native infrastructure for more than six years and has worked on various large-scale distributed systems, including cross datacenter workloads scheduler, multi-cloud management system, and distributed coordination service. In addition, he is a regular speaker presenting at various academic and industry conferences.


🔶 Michael Bowen, Sr. Principal Engineer at BlackRock

Michael is an Open Source advocate, evangelist, and engineer on the Platform Engineering team at BlackRock. Heavily focused on cloud-native solutions targeting container workloads, Kubernetes orchestration, and application delivery to consumers.