Predicting Disk Scheduling Performance with Virtual Machines

  • Robert Geist
  • Zachary H. Jones
  • James Westall
Part of the Lecture Notes in Computer Science book series (LNCS, volume 6821)


A method for predicting the performance of disk scheduling algorithms on real machines using only their performance on virtual machines is suggested. The method uses a dynamically loaded kernel intercept probe (iprobe) to adjust low-level virtual device timing to match that of a simple model derived from the real device. An example is provided in which the performance of a newly proposed disk scheduling algorithm is compared with that of standard Linux algorithms. The advantage of the proposed method is that reasonable performance predictions may be made without dedicated measurement platforms and with only relatively limited knowledge of the performance characteristics of the targeted devices.


Virtual Machine Service Time Virtual Machine Monitor Virtual System Sleep Interval 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


  1. 1.
    Creasy, R.: The origin of the VM/370 time-sharing system. IBM Journal of Research & Development 25, 483–490 (1981)CrossRefGoogle Scholar
  2. 2.
    Popek, G.J., Goldberg, R.P.: Formal requirements for virtualizable third generation architectures. Commun. ACM 17, 412–421 (1974)CrossRefzbMATHMathSciNetGoogle Scholar
  3. 3.
    Robin, J.S., Irvine, C.E.: Analysis of the Intel®Pentium’sTMability to support a secure virtual machine monitor. In: Proc. of the 9th Conf. on USENIX Security Symposium (SSYM 2000), pp. 10–10. USENIX Assoc., Berkeley (2000)Google Scholar
  4. 4.
    VMWare, Inc.: Understanding full virtualization, paravirtualization, and hardware assist (2007),
  5. 5.
    Barham, P., Dragovic, B., Fraser, K., Hand, S., Harris, T., Ho, A., Neugebauery, R., Pratt, I., Warfield, A.: Xen and the art of virtualization. In: Proc. 19th ACM Symp. on Operating System Principles, Bolton Landing, New York, pp. 164–177 (2003)Google Scholar
  6. 6.
    Kivity, A., Kamay, Y., Laor, D., Lublin, U., Liguori, A.: kvm: the Linux virtual machine monitor. In: Proceedings of the Linux Symposium, Ottawa, Ontario, Canada, pp. 225–230 (2007)Google Scholar
  7. 7.
    Geist, R., Jones, Z., Westall, J.: Virtualizing high-performance graphics cards for driver design and development. In: Proc. 19th Annual Int. Conf. of the IBM Centers for Advanced Studies (CASCON 2009), Toronto, Ontario, Canada (2009)Google Scholar
  8. 8.
    Geist, R., Jones, Z., Westall, J.: Virtualization of an advanced course in operating systems. In: Proc. 3rd Int. Conf. on the Virtual Computing Initiative (ICVCI3), Raleigh, North Carolina (2009)Google Scholar
  9. 9.
    Jones, Z.H.: A Framework for Virtual Device Driver Development and Virtual Device-Based Performance Modeling. PhD thesis, Clemson University (2010)Google Scholar
  10. 10.
    Coffman, E., Hofri, M.: On the expected performance of scanning disks. SIAM J. on Computing 11, 60–70 (1982)CrossRefzbMATHMathSciNetGoogle Scholar
  11. 11.
    Coffman Jr., E.G., Gilbert, E.N.: Polling and greedy servers on a line. Queueing Syst. 2, 115–145 (1987)Google Scholar
  12. 12.
    Geist, R., Daniel, S.: A continuum of disk scheduling algorithms. ACM TOCS 5, 77–92 (1987)CrossRefGoogle Scholar
  13. 13.
    Worthington, B.L., Ganger, G.R., Patt, Y.N.: Scheduling algorithms for modern disk drives. In: Proceedings of the 1994 ACM SIGMETRICS Conference on Measurement and Modeling of Computer Systems, Nashville, TN, USA, pp. 241–251 (1994)Google Scholar
  14. 14.
    Pratt, S., Heger, D.A.: Workload dependent performance evaluation of the Linux 2.6 i/o schedulers. In: Proc. of the Linux Symposium, Ottawa, Ontario, vol. 2, pp. 425–448 (2004)Google Scholar
  15. 15.
    Geist, R., Ross, R.: Disk scheduling revisited: Can o(n 2) algorithms compete? In: Proc. of the 35th Annual ACM SE Conf., Murfreesboro, Tennessee, pp. 51–56 (1997)Google Scholar
  16. 16.
    Bellman, R.E.: Dynamic Programming. Dover Publications (2003) (incorporated) Google Scholar
  17. 17.
    Geist, R., Steele, J., Westall, J.: Enhancing webserver performance through the use of a drop-in, statically optimal, disk scheduler. In: Proc. of the 31st Annual Int. Conf. of the Computer Measurement Group (CMG 2005), Orlando, Florida, pp. 697–706 (2005)Google Scholar
  18. 18.
    Iyer, S., Druschel, P.: Anticipatory scheduling: a disk scheduling framework to overcome deceptive idleness in synchronous i/o. In: SOSP 2001: Proceedings of the Eighteenth ACM Symposium on Operating Systems Principles, pp. 117–130. ACM, New York (2001)CrossRefGoogle Scholar
  19. 19.
    Barford, P., Crovella, M.: Generating representative web workloads for network and server performance evaluation. In: Proc. ACM SIGMETRICS Measurement and Modeling of Computer Systems, pp. 151–160 (1998)Google Scholar
  20. 20.
    Douglas Gilbert: sdparm utility 1.03 (2008),
  21. 21.
    Qian, J., Meyers, C.R., Wang, A.I.A.: A Linux implementation validation of track-aligned extents and track-aligned raids. In: USENIX Annual Technical Conference, Boston, MA, pp. 261–266 (2008)Google Scholar
  22. 22.
    Seagate Technology LLC: Product Manual Cheetah 15K.4 SCSI. Pub. no. 100220456, rev. d edn., Scotts Valley, CA (2005)Google Scholar

Copyright information

© IFIP International Federation for Information Processing 2011

Authors and Affiliations

  • Robert Geist
    • 1
  • Zachary H. Jones
    • 1
  • James Westall
    • 1
  1. 1.Clemson UniversityUSA

Personalised recommendations