CS 562: Virtual Machines

Overview

Virtual machines have transformed the ways in which we build, manage, and interact with modern computer systems. A great deal of the network packets that you send as you sit at your computer are handled by virtual machines that may move to another side of the world in a matter of hours. If you have an Android phone, every application you launch executes in a virtual machine. Have you written a Java or Python program lately? It couldn't run without a virtual machine sitting underneath it. VMs have enabled new development practices for production applications, mobile applications, web applications, and operating system kernels. They allow datacenter operators to more efficiently provision hardware resources, saving money and energy. They allow service providers to quickly react to failures in an efficient and clean way. They enable the construction of portable and platform-agnostic programming languages by decoupling applications from the hardware on which they run. While VMs have been around since the early 1970s, modern developments, from cloudlets to containers, are only increasing the utility of virtualization technologies. One can expect that their importance and relevance will only increase.

This means that a basic understanding of the technological foundations underlying virtual machines should be part of any computer scientist's repertoire. This course will draw back the curtains and expose the magic that makes various types of VMs work. By the end of the course, you will have a deep understanding of hypervisors, system virtualization, machine emulation, language virtual machines, binary translators, virtual resource management, and more. You will gain exposure to a real-world hypervisor code-base. Furthermore, you will actually build a virtual machine and develop an intuition for using VMs to solve problems. The course will involve lectures, written assignments, involved programming projects, and discussions of foundational research papers.

Course Goals

The goals fo this course will be for you to develop a deep understanding of various types of virtualization techniques, their advantages and disadvantage, and to be able to apply them in a practical setting. You will be able to build basic VM constructs and understand how to evaluate them. You will therefore be expected to strengthen your system programming skills. You will also learn about new and upcoming technologies related to virtualization.

In general, you will learn about the following topics (and potentially others):

• General resource virtualization: virtual memory, virtualized CPUs, OS abstractions
• Basic Virtual Machines and historical context. Formal requirements for virtual machines
• Emulation: interpretation and binary translation
• Language VMs and Process VMs: JVM, Python, .NET CLR, Android's Dalvik JVM, VCODE, JIT compilation
• Full system virtualization: Type I & II VMMs, paravirtualization, device virtualization
• Virtualization in practice: virtual resource management, live migration, memory ballooning, fault tolerance
• Network virtualization: virtual overlays, bridged networking, NAT, software-defined networks
• Virtual storage: storage virtualization, virtual disk formats, copy-on-write formats
• Advanced topics in virtualization: hardware assisted virtualization, overheads of virtualization, performant VMs, interrupt virtualization, self-virtualizing devices, security, introspection, the semantic gap
• Current and future directions: VMMs as microkernels, Unikernels, chroot, jails, containers, cloudlets, virtual appliances

Communication

We will be primarily using Piazza as a course communication mechanism. If you have an issue or question that is not strictly private (especially one you think would benefit everyone were it answered), please use Piazza as your first resource. The instructor and your fellow classmates will be there to help. Note that you can also post anonymously if you so choose.

---

---

Books

The following book is the only required textbook for this course. If you plan on pursuing computer systems seriously, it is a great book to have as a reference:

Development Environment

For all projects I will provide you with a Vagrant configuration which you can use to spawn a virtual machine to do your work on. To use Vagrant, you'll need a VMM installed on your machine. VirtualBox and libvirt (built on QEMU/kvm) are two free options. VMware Workstation or Fusion for Mac will work as well. If you're on a Windows box, HyperV is another option. See here for getting started with Vagrant VMs.

Other Useful Links and Resources

This is a list of other resources that you might find useful for this class and for doing work in the systems area in general. Feel free to peruse them at your own convenience.

Links

• Nice tutorial for learning Git.
• Emulators for various historical machines
• MS BASIC for 6502
• Interrupt logic on the 6502
• 6502 on the cloud
• Visual 6502
• 6502 Instruction Set Reference
• 6502 Programmer's Manual