SlideShare a Scribd company logo
1 of 1
Measurement of TCP/IP Performance
under UNIX Operating System :
To analyze performance of the TCP/IP under UNIX
operating system, we focused on the performance of
the data exchange during a regular data transfer once
a connection is established. Performance is measured
in terms of cycle counts, executed instruction counts,
data read counts, and data write counts. Since the
Cycle counter in the Pentium processor continues to
increment even when the machine is halted, also since it
increments when the processor is switched to run other
processes and interrupts, it is very hard to measure
the precise execution time of the program block being
executed. To ensure the measurement as accurate as
possible, we repeated the same measure several times
to take out such results which are far off the standard
deviation.
We should note that the perturbation of measurement
is not due to the measurement errors hit due
to the running conditions of the Pentium processor.
For example, depending on instruction and data cache
miss ratios, the cycle counts vary greatly. Also, the
instruction counts are much affected by the internal
pipelines state of the processor. Therefore, our measurement
data are taken only when cache miss rates
and pipelines states are within some specified ranges,
discarding ones far out of ranges. By doing so, we can
have an idea of relative execution time of the program
under measure using the cycle counts in conjunction
with the event counts such as instruction counts and
data read/write counts.
Ye also did not consider functions which are not
directly related with the regular data transfer such as
urgent segment, errror messages, etc.. Thus, the overall
performance which might be drawn from our empirical
results should imply upper bounds on performance

More Related Content

What's hot

Transport triggered architecture
Transport triggered architectureTransport triggered architecture
Transport triggered architecture
Gichelle Amon
 
process control blockPcb
process control blockPcbprocess control blockPcb
process control blockPcb
Parul Shree
 
Centralized decentralized discrete control systems
Centralized decentralized discrete control systemsCentralized decentralized discrete control systems
Centralized decentralized discrete control systems
Ajnas KC
 

What's hot (18)

The process states
The process statesThe process states
The process states
 
11 instruction sets addressing modes
11  instruction sets addressing modes 11  instruction sets addressing modes
11 instruction sets addressing modes
 
Manish tripathi-itim-incident-management
Manish tripathi-itim-incident-managementManish tripathi-itim-incident-management
Manish tripathi-itim-incident-management
 
14 superscalar
14 superscalar14 superscalar
14 superscalar
 
Homework solutionsch9
Homework solutionsch9Homework solutionsch9
Homework solutionsch9
 
12 processor structure and function
12 processor structure and function12 processor structure and function
12 processor structure and function
 
Memory management basics
Memory management basicsMemory management basics
Memory management basics
 
Consistency protocols
Consistency protocolsConsistency protocols
Consistency protocols
 
13 risc
13 risc13 risc
13 risc
 
Client Centric Consistency Model
Client Centric Consistency ModelClient Centric Consistency Model
Client Centric Consistency Model
 
Transport triggered architecture
Transport triggered architectureTransport triggered architecture
Transport triggered architecture
 
process control blockPcb
process control blockPcbprocess control blockPcb
process control blockPcb
 
Centralized decentralized discrete control systems
Centralized decentralized discrete control systemsCentralized decentralized discrete control systems
Centralized decentralized discrete control systems
 
DATA TRANSFER SCHEMES OF 8085
DATA TRANSFER SCHEMES OF 8085DATA TRANSFER SCHEMES OF 8085
DATA TRANSFER SCHEMES OF 8085
 
Chapter01-rev.pptx
Chapter01-rev.pptxChapter01-rev.pptx
Chapter01-rev.pptx
 
RTOS APPLICATIONS
RTOS  APPLICATIONSRTOS  APPLICATIONS
RTOS APPLICATIONS
 
Process concept
Process conceptProcess concept
Process concept
 
What is Database Backup? The 3 Important Recovery Techniques from transaction...
What is Database Backup? The 3 Important Recovery Techniques from transaction...What is Database Backup? The 3 Important Recovery Techniques from transaction...
What is Database Backup? The 3 Important Recovery Techniques from transaction...
 

Viewers also liked (17)

Pnl educación financiera
Pnl educación financieraPnl educación financiera
Pnl educación financiera
 
Ac ficha de revisões m4
Ac ficha de revisões  m4Ac ficha de revisões  m4
Ac ficha de revisões m4
 
Dessert tart
Dessert tartDessert tart
Dessert tart
 
Final Bluff Schedule
Final Bluff ScheduleFinal Bluff Schedule
Final Bluff Schedule
 
Sinais e sintomas
Sinais e sintomasSinais e sintomas
Sinais e sintomas
 
El teletrabajo
El teletrabajoEl teletrabajo
El teletrabajo
 
Ind quadri420
Ind quadri420Ind quadri420
Ind quadri420
 
For the newbie
For the newbieFor the newbie
For the newbie
 
Hola
HolaHola
Hola
 
Pesquisa no campo e a justiça social
Pesquisa no campo e a justiça socialPesquisa no campo e a justiça social
Pesquisa no campo e a justiça social
 
Keira's choice of athlete
Keira's choice of athleteKeira's choice of athlete
Keira's choice of athlete
 
Familia de Aire Acondicionado Air Pro
Familia de Aire Acondicionado Air ProFamilia de Aire Acondicionado Air Pro
Familia de Aire Acondicionado Air Pro
 
Evangelista
EvangelistaEvangelista
Evangelista
 
RN 154 : reunion publique 10 juillet 19h Chambre d'Agriculture à Chartres
RN 154 : reunion publique 10 juillet 19h Chambre d'Agriculture à ChartresRN 154 : reunion publique 10 juillet 19h Chambre d'Agriculture à Chartres
RN 154 : reunion publique 10 juillet 19h Chambre d'Agriculture à Chartres
 
Maria cenaculo
Maria cenaculoMaria cenaculo
Maria cenaculo
 
CÉLIA SANTOS
CÉLIA SANTOSCÉLIA SANTOS
CÉLIA SANTOS
 
Call Sheet
Call SheetCall Sheet
Call Sheet
 

Similar to Measurement of tcp

Basic Structure of Computers: Functional Units, Basic Operational Concepts, B...
Basic Structure of Computers: Functional Units, Basic Operational Concepts, B...Basic Structure of Computers: Functional Units, Basic Operational Concepts, B...
Basic Structure of Computers: Functional Units, Basic Operational Concepts, B...
Abhishekn84
 
ROLE OF DIGITAL SIMULATION IN CONFIGURING NETWORK PARAMETERS
ROLE OF DIGITAL SIMULATION IN CONFIGURING NETWORK PARAMETERSROLE OF DIGITAL SIMULATION IN CONFIGURING NETWORK PARAMETERS
ROLE OF DIGITAL SIMULATION IN CONFIGURING NETWORK PARAMETERS
Deepak Shankar
 
Whitepaper: Where did my CPU go?
Whitepaper: Where did my CPU go?Whitepaper: Where did my CPU go?
Whitepaper: Where did my CPU go?
Kristofferson A
 

Similar to Measurement of tcp (20)

Connectivity Solutions with InduSoft Web Studio and Software toolbox.
Connectivity Solutions with InduSoft Web Studio and Software toolbox.Connectivity Solutions with InduSoft Web Studio and Software toolbox.
Connectivity Solutions with InduSoft Web Studio and Software toolbox.
 
Parallel machines flinkforward2017
Parallel machines flinkforward2017Parallel machines flinkforward2017
Parallel machines flinkforward2017
 
Lecture notes -001
Lecture notes -001Lecture notes -001
Lecture notes -001
 
Co m1-1
Co m1-1Co m1-1
Co m1-1
 
Basic Structure of Computers: Functional Units, Basic Operational Concepts, B...
Basic Structure of Computers: Functional Units, Basic Operational Concepts, B...Basic Structure of Computers: Functional Units, Basic Operational Concepts, B...
Basic Structure of Computers: Functional Units, Basic Operational Concepts, B...
 
chapter 1 -Basic Structure of Computers.pptx
chapter 1 -Basic Structure of Computers.pptxchapter 1 -Basic Structure of Computers.pptx
chapter 1 -Basic Structure of Computers.pptx
 
13009690.ppt
13009690.ppt13009690.ppt
13009690.ppt
 
Programmable logic controller
Programmable logic controllerProgrammable logic controller
Programmable logic controller
 
Assembly p1
Assembly p1Assembly p1
Assembly p1
 
Network architecture (protocols) by lanz lowen jay o. bognot
Network architecture (protocols) by lanz lowen jay o. bognotNetwork architecture (protocols) by lanz lowen jay o. bognot
Network architecture (protocols) by lanz lowen jay o. bognot
 
Computer architecture short note (version 8)
Computer architecture short note (version 8)Computer architecture short note (version 8)
Computer architecture short note (version 8)
 
Transport layer
Transport layerTransport layer
Transport layer
 
Control of Switching.pdf
Control of Switching.pdfControl of Switching.pdf
Control of Switching.pdf
 
Polling_Computer_Organisation.ppt
Polling_Computer_Organisation.pptPolling_Computer_Organisation.ppt
Polling_Computer_Organisation.ppt
 
ROLE OF DIGITAL SIMULATION IN CONFIGURING NETWORK PARAMETERS
ROLE OF DIGITAL SIMULATION IN CONFIGURING NETWORK PARAMETERSROLE OF DIGITAL SIMULATION IN CONFIGURING NETWORK PARAMETERS
ROLE OF DIGITAL SIMULATION IN CONFIGURING NETWORK PARAMETERS
 
Whitepaper: Where did my CPU go?
Whitepaper: Where did my CPU go?Whitepaper: Where did my CPU go?
Whitepaper: Where did my CPU go?
 
IRJET- Simulation Analysis of a New Startup Algorithm for TCP New Reno
IRJET- Simulation Analysis of a New Startup Algorithm for TCP New RenoIRJET- Simulation Analysis of a New Startup Algorithm for TCP New Reno
IRJET- Simulation Analysis of a New Startup Algorithm for TCP New Reno
 
DSP Processor
DSP Processor DSP Processor
DSP Processor
 
Traffic Simulator
Traffic SimulatorTraffic Simulator
Traffic Simulator
 
SWITCHING SYSTEM SOFTWARE
SWITCHING SYSTEM SOFTWARESWITCHING SYSTEM SOFTWARE
SWITCHING SYSTEM SOFTWARE
 

Measurement of tcp

  • 1. Measurement of TCP/IP Performance under UNIX Operating System : To analyze performance of the TCP/IP under UNIX operating system, we focused on the performance of the data exchange during a regular data transfer once a connection is established. Performance is measured in terms of cycle counts, executed instruction counts, data read counts, and data write counts. Since the Cycle counter in the Pentium processor continues to increment even when the machine is halted, also since it increments when the processor is switched to run other processes and interrupts, it is very hard to measure the precise execution time of the program block being executed. To ensure the measurement as accurate as possible, we repeated the same measure several times to take out such results which are far off the standard deviation. We should note that the perturbation of measurement is not due to the measurement errors hit due to the running conditions of the Pentium processor. For example, depending on instruction and data cache miss ratios, the cycle counts vary greatly. Also, the instruction counts are much affected by the internal pipelines state of the processor. Therefore, our measurement data are taken only when cache miss rates and pipelines states are within some specified ranges, discarding ones far out of ranges. By doing so, we can have an idea of relative execution time of the program under measure using the cycle counts in conjunction with the event counts such as instruction counts and data read/write counts. Ye also did not consider functions which are not directly related with the regular data transfer such as urgent segment, errror messages, etc.. Thus, the overall performance which might be drawn from our empirical results should imply upper bounds on performance