Skip to content

The test to follow after Non Parallel Processing

Notifications You must be signed in to change notification settings

XoulKool/P-Processing

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

6 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

P-Processing

The test to follow after Non Parallel Processing

Once again this test requires you to have the same five terminals as in the previous test - That is the Pica8, grnlntrn, server5, server6, and server7. Instructions for sshing into the terminal were given in the last tests, read me so please look at that. Again, remember to delete any flows that have been created by prior tests in the Pica8 by using the commands specified in the last two tests on the Pica8 terminal. Also, this test uses OpenFlow 1.0 so make sure the Pica8 switch is set for this version using the command specified in the last two tests.

Note that in this test the goal is to time the transfer of two large data files in parallel, that is sending two different files through the same port on the Pica8 switch at one time and measuring how much faster this is than round-robin scheduling transfers. Keep in mind your transfer files will already need to be created and the two listening servers must know the name of these files.

The first part of the test invloves starting up the two listening servers on servers 5 and 7. Do this on server 5 with command

python PServer1.py

and this command on server7

python PServer2.py

Now run simple_switch.py which is a default Layer 2 switch that comes with Ryu. [it will forward transfer files an put in the proper flows] Do this with command on grnlntrn server

ryu-manager --verbose simple_switch.py

Next run client file on server6 [after grnlntrn terminal says that it has moved to main mode] using command

python PClient.py

and now all terminals should be doing something, and you should see in roughly twenty seconds (if the files you are transferring are 1GB files) that booth files were given to the servers successfully. Now if you look at bparlog1.txt and bparlog2.txt we can see how quickly each fiile was transferred. We can then compare these times to the times displayed in the prior test and see that the overhead of adding and installing flows was substantial enough to prove that just regular prarallel file transfer is faster than a round-robin file transfer. This is kind of a complex sanity check on this switch, as this is how any commercial-grade switch should be behaving.

About

The test to follow after Non Parallel Processing

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages