Xen and XenServer Storage Performance
summary
Measurements show a host with three RAID0 volumes (Seagate 15k SAS, WD SSD, Intel DC SSD) behind a PERC H700 and Linux 3.11 yields dom0 latencies ~50–100µs while guests see several hundred µs and lower throughput with blkfront/blkback or user‑space tapdisk/qemu paths. Tracing isolates costs in grant mapping, event channels, context switches and copying. Persistent grants reduce grant mapping contention but increase copy time; indirect I/O and ring/segment limits (11 segments ≈44KB, ~32 slots) explain why larger requests and millions of IOPS need new paths.
AI-written summary · about
chapters (6)
- 0:00Start
- 1:16Where do Xen and XenServer stand?
- 3:20Current Implementation (Xen)
- 13:10Measurement Over Different Back Ends
- 19:03Persistent Grants Assessment
- 20:13New Ideas
related talks
40mAchieving the Ultimate Performance with KVM
43mComparing Performance of NVMe Hard Drives in KVM, Baremetal, and Docker
19mSpeeding Up VM’s I/O Sharing Host's io_uring Queues With Guests
26mAnalysis of AMD HW-assisted vIOMMU Implementation and Performance
40mBenchmarking: Repeatable & Comparable
21mNVMe Emulation Performance Optimization
15mThe Practice Method to Speed Up 10x Boot-up Time for Guest in Alibaba Cloud
25mKVM Dirty Page Tracking