The eBay story with Swift was born many years ago, as this object storage solution is architected for massive scaling and designed for large data access.
The eBay storage team enhanced Swift further to fit many in-house requests and to achieve high scale. Come join us for this presentation where we will discuss:
- Quality of Service (QoS), Rate Limiting
- Caching layer - pushed the concurrent download to 10x
- Persistent Storage across data centers
- Large number of Request Per Seconds
- Higher Durability, Reliability and Availability
- Multi-Tenancy / zone support - Serve multiple zones/ VPC from one cluster
We will also discuss operational challenges and how eBay solved them:
- Swift utilization at eBay
- Monitoring Swift (ELK)
- Applied Hardware optimization and monitoring
- Splitting Proxy-Account-Container-Object (PACO), into PAC-O
- Archival storage solution with Erasure Coding
- Security HTTPs/TLS 1.2 enforcements
- Swift Horizontal scaling for “massive” throughput
- Swift on Kubernetes
We will take you through interesting journey where team will share challenges with productionizing Swift and how they solved:
- Large Scale
- Securtiy
- Performance
- Availability and Durability
- Reliability
- Multi VPC / Security Zone support
- Observability and Monitoring
- Optimizations (HW and SW)
