Machine Setup

Hardware details

• Processor:

 Hardware Class: cpu
  Arch: X86-64
  Vendor: "AuthenticAMD"
  Model: 15.79.2 "AMD Athlon(tm) 64 Processor 3200+"
  Features: fpu,vme,de,pse,tsc,msr,pae,mce,cx8,apic,sep,mtrr,pge,mca,cmov,pat,pse36,clflush,mmx,
fxsr,sse,sse2,syscall,nx,mmxext,fxsr_opt,lm,3dnowext,3dnow,pni,cx16,lahf_lm
  Clock: 2009 MHz
  Cache: 512 kb
  Config Status: cfg=new, avail=yes, need=no, active=unknown

No virtualization instructions available.

• Ram:
• Giga-byte C51 Host Bridge
• 512 MB
• HD:
• Seagate Barracuda 7200.9 160GB S-ATA II 8MB (st3160811as)
• Giga-byte MCP51 Serial ATA Controller
• 160 GB
• Ethernet
• Nvidia Giga-byte MCP51 Ethernet Controller
• Intel PRO/1000 MT Desktop Adapter
• D-Link DFE-538TX 10/100 Ethernet Adapter

Host system

• Debian etch etch-64
• Use of lvm
• Use of ext3 as filesystem

Virtual guest systems

• Debian etch : debootstrap
• GNU GCC compiler
• Apache webserver

Xen install

• Kernel :
• Packages :
• Config files
• machine config
• network config
• Setup DomU system

Vserver install

Benchmark setup

• Covered points in the benchmark
• CPU utilization
• Inter-process communication
• Hard disk access
• Network performance

• More info about linux benchmarks : http://lbs.sourceforge.net/
• Used benchmark software
• Unixbench -- CPU and IPC benchmarks
• Siege -- Performance of HTTP request

UnixBench 4.1.0

• http://www.tux.org/pub/tux/benchmarks/System/unixbench/
• Testsetup
• Run on a non-virtualized OS (dom0)
• Insert the results as a base value
• Subsequent tests for guests in virtual environments are referring to this value.

   Most Unixench microbenchmarks execute a number of iterations and report results for a particular benchmark as
an average of those iterations. In order to provide a greater degree of normalization, we chose to run the Unixbench
suite three times each in various user modes (i.e. – root user unenforced, root user enforced, regular user unenforced,
etc) on both testbed systems. The results from the three runs have been compiled and we report the average of these
runs. The Unixbench benchmark suite includes the following microbenchmarks that we selected to include in our
report:
          Whetstone- assignment, addition, subtraction & multiplication calculations that substitute datatypes for
          numbers (register, short, int, float, long, double and an empty loop)
          Dhrystone 2- manipulation of arrays, character strings, indirect addressing, and most of the non-floating
          point instructions that might be found in an application program. It also includes conditional operations and
          other common program flow controls.
          Execl- replacing a currently running process with a new process
         File copy 4K/1K/256B buffer sizes- captures the number of characters that can be copied within 10 seconds
         based on varying buffer sizes
         Pipe throughput- a single process opens a pipe (an inter-process communications channel) to itself and
         spins a megabyte around a short loop.
         Context-switching- a test program spawns a child process with which it carries on a bi-directional pipe
         conversation.
         Process creation- a test that creates a child process which immediately dies after its own fork(). The process
         is repeated over and over.
         Shell scripts- a shell script that is run by 1, 2, 4, and 8 concurrent processes. The script consists of an
         alphabetic sort one file to another; taking the octal dump of the result and doing a numeric sort to a third
         file; running grep on the result of the alphabetic sort file; “tee”ing the result to a file and to wc (wordcount);
         writing the final result to a file; and removing all of the resulting files.
         Syscall- this test evaluates the time required to do iterations of dup(), close(), getpid(),getuid(), and umask()
         calls.
As described above, the averaged results yielded by UnixBench are included in Figure 1. While a detailed
interpretation of these results requires understanding of what metric each benchmarks reports, higher index numbers
indicate better performance.

• Or the linux benchmarking howto: tldp.org/HOWTO/Benchmarking-HOWTO.html

• What: measures overall Unix performance. This test will exercice the file I/O and kernel
  multitasking performance.
• I have discarded all arithmetic test results, keeping only the system−related test results.
• Test procedure: make with −O2. Execute with ./Run −1 (run each test once). You will find the
  results in the ./results/report file. Calculate the geometric mean of the EXECL THROUGHPUT,
  FILECOPY 1, 2, 3, PIPE THROUGHPUT, PIPE−BASED CONTEXT SWITCHING, PROCESS
  CREATION, SHELL SCRIPTS and SYSTEM CALL OVERHEAD indexes.
• Results: a system index.