41 phone booth packing phone booth packing is a

This preview shows page 9 - 11 out of 13 pages.

4.1 Phone Booth Packing Phone booth packing is a contest of cramming as many people as possible into a booth designed for one person. We have a similar benchmark to deter- mine the upper limit of packing concurrently active VMs onto a single physical host. It essentially is a footprint benchmark measured in terms of available disk and memory space. We evaluate the virtual- ization approaches in two dimensions: guaranteed reservation vs. overbooking. For guaranteed reser- vation the underlying VMM admits only the num- ber of VMs that it can guarantee the availability of a reserved amount of resources. For overbooking, the VMM admits more VMs and services them on 9
Image of page 9

Subscribe to view the full document.

a best effort basis. The assumption being that in the common case VMs will utilize far less resources than their maximim limits permit. With guaranteed reservations (i.e., 100MB mem- ory and 1GB disk space) both systems scale simi- larly, as the limiting factor is available memory and disk space—you can only cut a pie so often. The real challenge is in reducing the footprint of individ- ual VMs when overbooking the physical resources. In general, for hypervisor-based systems there is an inherent overcommittment of virtual mem- ory pages and disk blocks to each VM. To reduce the memory footprint, such systems need to reduce duplicate state–introduced by each VM running a Guest OS and its applications– after the fact . While there are several proposed techniques to reduce du- plicate memory pages [18, 17], each comes with its own drawbacks. For example, content-based page sharing techniques [18] compute hash val- ues by scanning pages to identify duplicate pages and then uses CoW virtual memory mechanisms to reduce the memory footprint. Similarly, the delta-virtualization techniques described by Vrable et. al. [17] are very specialized for short-running applications (operating at the level of seconds or minutes)—not the usage scenarios we are investi- gating. These memory footprint optimization techniques could also be applied to paenevirtualized systems to reduce memory footprint. However, their pri- mary benefit is to reduce the text segments of com- mon applications. When properly configured, this is something that paenevirtualized systems already do inherently. For example, on PlanetLab and other Linux Vserver-based systems (such as Lycos Eu- rope), the filesystem is shared using a file-level CoW scheme. Many of the common executables and shared libraries are thereby shared on disk be- tween VMs (vservers). More importantly, their in- odes are the same and file paths appear the same across VMs. This setup lets the dynamic loader map them into virtual memory such that they are shared read-only across processes, even those that run in separate VMs. Squeezing as many VM’s filesystem onto a disk as possible, though, is more an exercise in frugal- ity and ease of use rather than in reducing dupli- cate state. The ideal case for Xen might be to use file-backed virtual block devices (fVBD). A fVBD can be configured as a sparse file, which lets one nicely overbook the amount of disk space. However, the current implementation of the backing store for fVBDs caches I/O in the host domain, i.e., it be-
Image of page 10
Image of page 11
You've reached the end of this preview.
  • Spring '12
  • GwangS.Jung
  • virtual machine, VMs, Linux kernel, Xen, namespace isolation

{[ snackBarMessage ]}

What students are saying

  • Left Quote Icon

    As a current student on this bumpy collegiate pathway, I stumbled upon Course Hero, where I can find study resources for nearly all my courses, get online help from tutors 24/7, and even share my old projects, papers, and lecture notes with other students.

    Student Picture

    Kiran Temple University Fox School of Business ‘17, Course Hero Intern

  • Left Quote Icon

    I cannot even describe how much Course Hero helped me this summer. It’s truly become something I can always rely on and help me. In the end, I was not only able to survive summer classes, but I was able to thrive thanks to Course Hero.

    Student Picture

    Dana University of Pennsylvania ‘17, Course Hero Intern

  • Left Quote Icon

    The ability to access any university’s resources through Course Hero proved invaluable in my case. I was behind on Tulane coursework and actually used UCLA’s materials to help me move forward and get everything together on time.

    Student Picture

    Jill Tulane University ‘16, Course Hero Intern