To run vgrid you need one or more computers with a running java runtime environment (Version 1.4+). You get a free java runtime for many different platforms from Sun Microsystems.
All computers should be connected to the server via TCP/IP. This is usually the case in a LAN.
One of the computers has to be the vgrid server. You need a servlet container being installed on that machine. I recommend tomcat from the "jakarta project. It is free (Open Source) and well tested im many projects around the world. Installing tomcat is very easy if you follow the instructions provided by the tomcat developer team.
You should test the prerequisites before you proceed. Type 'java -version' in a command line window of any of your computers. This should give you information about the version of the installed java runtime. The version should be 1.4 or above. Test the servlet container by starting the server. On a browser you should be capable to visit the welcome page of the container. For tomcat the welcome page can usually be accessed at 'http://localhost:8080/'.
After you have fulfilled the prerequisites on each of your computers you must deploy the vgrid server application on the computer where you installed the servlet container before. This can be done by copying the vgrid.war file into the server's webapps directory. After restarting the server the vgrid server page (http://NameOfYourServer:8080/vgrid) should be accessible in your internet browser. This page should be accessible from all the computers you want to use as part of the grid.
TIP: The server does not necessarily have to be the most powerful computer in the network. If it is not capable to handle the network traffic you may choose another computer later.
In order to evaluate tasks you need plodders on as many computers as possible.
The easiest way to start a plodder is to use java webstart. The vgrid server page contains a link that enables you to start a GUI plodder on any machine by just clicking the link on the vgrid server's welcome page. When the plodder is started it will immediately execute tasks from the server. If no tasks can be found there, it will request for unfinished tasks in regular intervals (3-4s)
Another way to start a plodder is to download the latest vgrid version from the vgrid download page. Expand the vgrid_plodder_x_x.zip file anywhere on your computer. It contains two .bat files that start a plodder on that machine. One runs with a swing GUI, the other one without GUI in a command line window. To tell the plodder where to find the server you have to edit the vgrid properties file before starting the plodder.
Plodder properties:
| name | description |
|---|---|
| vgrid.httpproxy.host | Your computer's name. Sample values: localhost, server, baghira, ... |
| vgrid.httpproxy.port | The port on which the server is listening. Sample values: 8080, 80, '', ... |
| vgrid.httpproxy.app | The name under which the vgrid server is deployed. Usually vgrid. |
| vgrid.httpproxy.nulltask_timeout | The time in millisecond the plodder is sleeping if it gets no connection from the server. If a connection can be established the plodder gets a null task from the server that controls the sleep time. In that case the nulltask_timeout value is ignored. Sample values: 1000, 3000, 5000, ... |
For Linux you may use the .bat files by just renaming them to .sh. I tried to avoid all platform specific syntax. Beware that this feature is not tested.
In order to learn how to implement your own tasks and how to dispense them within vgrid refer to the sample projects
Sample Projects:
| Gridaligner | Implementation of the Smith-Waterman algorithm for biological local pairwise sequence alignment. | GO |
| Gridwodka | Genetic algorithm to breed soda robots in a distributed environment. | NOT YET IMPLEMENTED |