Virtualization Series
A guided tour of traditional machine virtualization, organized as ten chapters from foundations through performance, plus per-system case studies grounded in source-code reading.
Virtualization Series
Virtualization Systems — Xen
Canonical paravirtualizing Type-1 hypervisor. PV/HVM/PVH modes contrasted side by side across CPU, memory, I/O, cross-domain communication, and VM management. The disaggregated shape in production form.
Virtualization Series 00 — History
A historical preface to the survey, organized into five eras from IBM CP-40 (1964) through modern microVMs and language-isolated systems. Anchors each era in the papers that invented the techniques later chapters treat as ambient.
Virtualization Series 01 — Foundations
What virtualization is, why it is used, and the Popek–Goldberg condition for virtualizability. Sets up the trap-and-emulate discipline that the rest of the survey treats as ambient context.
Virtualization Series 02 — Taxonomy
The four largely-independent axes of VMM design: placement (Type 1 vs Type 2), guest interface (full vs paravirtual), hardware support (VT-x, EPT, IOMMU), and isolation boundary (hardware, software, language).
Virtualization Series 03 — Hypervisor Architecture
The recurring component set inside a VMM (vCPU, memory, device, interrupt/timer, exit handler, control plane) and the three architectural shapes (monolithic, hosted, disaggregated) they organize into.
Virtualization Series 04 — CPU Virtualization
How guest code actually runs on a physical CPU: trap-and-emulate, binary translation, paravirtualization, and hardware-assisted virtualization (VT-x, AMD-V). Plus vCPU scheduling, gang scheduling, and the exit-cost story.
Virtualization Series 05 — Memory Virtualization
From shadow page tables to nested paging (EPT, NPT), with the overcommit toolbox: demand allocation, ballooning, content-based page sharing, hypervisor swapping, idle-memory taxation.
Virtualization Series 06 — I/O Virtualization
Full device emulation vs paravirtual (virtio + vhost) vs direct assignment (SR-IOV). The order-of-magnitude performance spread, the role of IOMMU, APICv, and posted interrupts, and how all three coexist in production.
Virtualization Series 07 — Cross-Domain Communication
Hypercalls, microkernel IPC, shared-memory rings, grant tables, capabilities — the substrate every non-monolithic VMM is built on. Cost anatomy of boundary crossings and why disaggregation lives or dies by it.