The Road to Stable and Fast Disks in Yandex Cloud
summary
Yandex Cloud storage team describes challenges of making block and networked disk storage both fast and reliable for cloud customers. They cover technologies and approaches for stability, performance and scalability of disk I/O in cloud environments, including adapting storage behavior to client load and operational practices used in Yandex Cloud.
AI-written summary · about
chapters (16)
- 0:00Introduction
- 1:09Cloud architecture and virtualization
- 5:00Virtual disk standards
- 6:25What is virtio and its benefits
- 8:15Virtio in QEMU
- 11:02Old solution problems
- 13:50New approach requirements
- 15:14Choosing VHost User
- 16:18How VHost User works
- 19:43Local and distributed disks
- 21:48Meeting requirements
- 23:13VHost User library
- 24:57Optimizations and benchmarks
- 26:30Plans and future development
- 27:33Summary
- 29:04Q&A
related talks
29mDC-Store: Eliminating Noisy Neighbor Containers using Deterministic I/O Performance
48mSoftware Defined Storage the Linux way
14mZNS+: Advanced Zoned Namespace Interface for Supporting In-Storage Zone Compaction
24mxadfs: Inverted DAX
17mHow advancements in storage are triggering innovations in open-source and vice-versa
23mImproving data placement for Zoned Linux File systems
21mSpool: Reliable Virtualized NVMe Storage Pool in Public Cloud Infrastructure
52mPeeling Back the Layers of Storage