Preventing Performance Degradation on Operating System Reboots

Slides:



Advertisements
Similar presentations
Difference Engine: Harnessing Memory Redundancy in Virtual Machines by Diwaker Gupta et al. presented by Jonathan Berkhahn.
Advertisements

G Robert Grimm New York University Virtual Memory.
Fast and Safe Performance Recovery on OS Reboot Kenichi Kourai Kyushu Institute of Technology.
A Fast Rejuvenation Technique for Server Consolidation with Virtual Machines Kenichi Kourai Shigeru Chiba Tokyo Institute of Technology.
Virtualization in HPC Minesh Joshi CSC 469 Dr. Box Feb 1, 2012.
Kenichi Kourai (Kyushu Institute of Technology) Takeshi Azumi (Tokyo Institute of Technology) Shigeru Chiba (Tokyo University) A Self-protection Mechanism.
CSE506: Operating Systems Block Cache. CSE506: Operating Systems Address Space Abstraction Given a file, which physical pages store its data? Each file.
Database Systems on Virtual Machines: How Much Do We Lose? Kristin Travis March 2, 2011.
A Secure System-wide Process Scheduling across Virtual Machines Hidekazu Tadokoro (Tokyo Institute of Technology) Kenichi Kourai (Kyushu Institute of Technology)
CSCE 212 Chapter 7 Memory Hierarchy Instructor: Jason D. Bakos.
Chapter 13 – File and Database Systems
CS 333 Introduction to Operating Systems Class 18 - File System Performance Jonathan Walpole Computer Science Portland State University.
CacheMind: Fast Performance Recovery Using a Virtual Machine Monitor Kenichi Kourai Kyushu Institute of Technology, Japan.
Fast and Correct Performance Recovery of Operating Systems Using a Virtual Machine Monitor Kenichi Kourai Kyushu Institute of Technology, Japan.
Tanenbaum 8.3 See references
Zen and the Art of Virtualization Paul Barham, et al. University of Cambridge, Microsoft Research Cambridge Published by ACM SOSP’03 Presented by Tina.
HyperSpector: Virtual Distributed Monitoring Environments for Secure Intrusion Detection Kenichi Kourai Shigeru Chiba Tokyo Institute of Technology.
Computer Architecture Lecture 28 Fasih ur Rehman.
Kenichi Kourai (Kyushu Institute of Technology) Takuya Nagata (Kyushu Institute of Technology) A Secure Framework for Monitoring Operating Systems Using.
Benefits: Increased server utilization Reduced IT TCO Improved IT agility.
CSE 451: Operating Systems Section 10 Project 3 wrap-up, final exam review.
Zero-copy Migration for Lightweight Software Rejuvenation of Virtualized Systems Kenichi Kourai Hiroki Ooba Kyushu Institute of Technology.
Our work on virtualization Chen Haogang, Wang Xiaolin {hchen, Institute of Network and Information Systems School of Electrical Engineering.
CS533 Concepts of Operating Systems Jonathan Walpole.
1 File Systems: Consistency Issues. 2 File Systems: Consistency Issues File systems maintains many data structures  Free list/bit vector  Directories.
 Virtual machine systems: simulators for multiple copies of a machine on itself.  Virtual machine (VM): the simulated machine.  Virtual machine monitor.
By Teacher Asma Aleisa Year 1433 H.   Goals of memory management  To provide a convenient abstraction for programming.  To allocate scarce memory.
Midterm Meeting Pete Bohman, Adam Kunk, Erik Shaw.
Lecture 11 Page 1 CS 111 Online Working Sets Give each running process an allocation of page frames matched to its needs How do we know what its needs.
CS333 Intro to Operating Systems Jonathan Walpole.
4.3 Virtual Memory. Virtual memory  Want to run programs (code+stack+data) larger than available memory.  Overlays programmer divides program into pieces.
Introduction: Memory Management 2 Ideally programmers want memory that is large fast non volatile Memory hierarchy small amount of fast, expensive memory.
Operating Systems ECE344 Ashvin Goel ECE University of Toronto Demand Paging.
UNIX & Windows NT Name: Jing Bai ID: Date:8/28/00.
Multilevel Caches Microprocessors are getting faster and including a small high speed cache on the same chip.
Protecting The Kernel Data through Virtualization Technology BY VENKATA SAI PUNDAMALLI id :
LINUX 15 : Memory Management
The Memory Hierarchy Lecture 31 20/07/2009Lecture 31_CA&O_Engr. Umbreen Sabir.
1 Contents Memory types & memory hierarchy Virtual memory (VM) Page replacement algorithms in case of VM.
Taeho Kgil, Trevor Mudge Advanced Computer Architecture Laboratory The University of Michigan Ann Arbor, USA CASES’06.
Virtual Machines (part 2) CPS210 Spring Papers  Xen and the Art of Virtualization  Paul Barham  ReVirt: Enabling Intrusion Analysis through Virtual.
Split Migration of Large Memory Virtual Machines
Translation Lookaside Buffer
Lesson Objectives Aims Key Words Paging, Segmentation, Virtual Memory
Memory Management Virtual Memory.
Jonathan Walpole Computer Science Portland State University
Virtual Memory Chapter 7.4.
Presented by Mike Marty
Memory COMPUTER ARCHITECTURE
Windows 95 & 98 Steve Boyle Mike Forster Maggie Hamill Nancy O’Brien.
Section 9: Virtual Memory (VM)
CS 704 Advanced Computer Architecture
Kenichi Kourai Hiroki Ooba Kyushu Institute of Technology, Japan
Virtual Memory © 2004, D. J. Foreman.
Introduction of Week 3 Assignment Discussion
I'm Kenichi Kourai from Kyushu Institute of Technology.
OS Virtualization.
Sho Kawahara and Kenichi Kourai Kyushu Institute of Technology, Japan
I'm Kenichi Kourai from Kyushu Institute of Technology.
Lecture 29: Virtual Memory-Address Translation
Translation Lookaside Buffer
Contents Memory types & memory hierarchy Virtual memory (VM)
Virtual Memory: Working Sets
Virtual Memory © 2004, D. J. Foreman.
Operating Systems: Internals and Design Principles, 6/E
System Virtualization
4.3 Virtual Memory.
Kenichi Kourai Kyushu Institute of Technology
Low-cost and Fast Failure Recovery Using In-VM Containers in Clouds
Efficient Migration of Large-memory VMs Using Private Virtual Memory
Presentation transcript:

Preventing Performance Degradation on Operating System Reboots Kenichi Kourai Tokyo Institute of Technology

Motivation OS reboot is not avoidable Many bugs in OSes Simple method for software rejuvenation Performance degradation after OS reboot The file cache is lost Disk access conflicts between virtual machines (VMs) VM VM OS reboot is not avoidable because operating systems have many bugs. When an operating system crashes or slows down critically, it must be rebooted. Applying patches to fix bugs often needs OS reboot. Furthermore, a technique called software rejuvenation has been proposed to prevent problems by software aging, but the simple method for OS rejuvenation is also OS reboot. Just after an operating system is rebooted, its performance is degraded. The primary cause is that the file cache on memory is lost and the cache miss rate increases. In addition, recently, multiple servers are consolidated into one physical machine with virtual machines. Under such server consolidation, disk access caused by cache miss conflicts between virtual machines and degrades the performance each other. shared physical disk

Warm-cache Reboot A new reboot mechanism to prevent the performance degradation Enables an OS to reuse the file cache Preserves the integrity of the file cache Our claim The file cache does not need to be discarded if its integrity is preserved The purpose of OS reboot is to initialize its internal state or to update its components To prevent such performance degradation after OS reboot, we propose a new reboot mechanism called the warm-cache reboot. This mechanism enables an operating system to be rebooted without losing the file cache and to reuse the file cache. The warm-cache reboot also preserves the integrity of the file cache. Our claim is that the file cache doesn’t need to be discarded if its integrity is preserved. The purpose of OS reboot is to initialize its internal state like data structures or to update its components, not to erase the file cache.

Preserving the File Cache The warm-cache reboot preserves cache pages on memory during OS reboot The VMM allocates the same memory to VMs The OS maintains a cache-mapping table cache-mapping table The warm-cache reboot preserves cache pages on memory during OS reboot. To achieve this, the virtual machine monitor allocates the same physical memory to rebooted virtual machines. In addition, the operating system maintains a table called cache-mapping table. This table is a hash table whose keys are a tuple of a device number, an inode number, and a file offset, and whose value is a frame number assigned to a cache page. When the operating system is rebooted, it first obtains the address of the cache-mapping table from the VMM and reserves its memory area. Then, it reserves all cache pages by traversing the table. cache page VM OS virtual machine monitor (VMM)

Protecting the File Cache Cache pages are write-protected The VMM reads file blocks into a protected cache page The VMM maintains a reuse flag write-protected To prevent cache pages from being corrupted by OS crashes, they are write-protected by the OS. In our system, the VMM reads file blocks from a disk into a protected cache page so that the contents are not corrupted during the file read. A cache page is write-protected in a VM, but the VMM can access it. To preserve the integrity of the file cache, the VMM maintains a reuse flag for each cache page. When the VMM reads file blocks into a cache page, it sets the reuse flag for the page. When the OS modifies the page, it unprotects the page and the VMM resets its reuse flag. If the reuse flag is not set, the cache page is not reused after OS reboot. VM OS set a reuse flag read virtual machine monitor (VMM) disk

Experiment Power test in DBT-3 Results VMM: Xen 3.0.0 OS: Linux 2.6.12 CPU: dual-core Opteron x2 Memory:12 GB Disk: Ultra SCSI Power test in DBT-3 To examine the performance just after OS reboot Results Warm-cache reboot No performance degradation Normal reboot Degraded by 67% To examine the performance just after OS reboot, we performed the power test in the DBT-3 benchmark. DBT-3 is a benchmark tool for databases and its power test measures the performance of the read access to databases. We measured the performance twice before OS reboot and twice after OS reboot. This figure shows the results. The pink bar is the result for the warm-cache reboot. The performance is not degraded just after OS reboot by reusing the file cache. On the other hand, in the normal reboot illustrated by the red bar, the performance is degraded by 67% due to cache misses. before after

Conclusion We proposed the warm-cache reboot Preserves the file cache during OS reboot Protects cache pages using the VMM Related work Rio file cache [Chen et al.'96] Preserves dirty file cache for reliability Warm-VM reboot [Kourai et al. '07] Preserves VMs during VMM reboot