Amazon EC2 Automation

Skills Needed • C# and .net • Mandatory: Extensive knowledge of Amazon EC2 and Amazon S3 • Optional: Extensive knowledge of TCP, UDP and modern Web Servers (in order to optimize the speed of http requests and report them back to the user pc) What the application should do The user works in a .net “Windows Forms?? application on his PC; this PC has a known IP address. On the users request, the App would then start a specified number of Amazon EC2 instances and install and start the same “worker?? program on each Amazon EC2 instance once it is started. The Amazon EC2 instances would run Windows XP/2003 Server/Windows 7. The worker programs then will start running their work load, and once finished, will report the result back to the user PC. The App will then graphically show the results to the user. At request of the user, all started EC2 instances should be stopped. Everything should be done in C# in .net. ## Deliverables **Skills Needed** · C# and .net · Mandatory: Extensive knowledge of Amazon EC2 and Amazon S3 · Optional: Extensive knowledge of TCP, UDP and modern Web Servers (in order to optimize the speed of http requests and report them back to the user pc) **What the application should do** The user works in a .net “Windows Forms?? application on his PC; this PC has a known IP address. On the users request, the App would then start a specified number of Amazon EC2 instances and install and start the same “worker?? program on each Amazon EC2 instance once it is started. The Amazon EC2 instances would run Windows XP/2003 Server/Windows 7. The worker programs then will start running their work load, and once finished, will report the result back to the user PC. The App will then graphically show the results to the user. At request of the user, all started EC2 instances should be stopped. Everything should be done in C# in .net. ** ** **The Worker Application:** Public class Worker { public Worker ( string ip_of_user_pc, int port_of_user_pc) { Open TCP socket, connect to user_pc DateTime started = [login to view URL]; Send to User PC start time Double result = DoWork(); DateTime ended = [login to view URL]; Send to User PC result and end time. } public double DoWork () { // In here we will put our current code; in essence it will be a few http requests, // parse the results, and report a double back } } **The Worker Application** Contains a form with the EC2 and S3 credentials. Button -> Upload Worker Image to S3 (This could take time, it then notifies the user once finished, or error message if there was an error) Combo - # of instances. Button Start Workers: This would then create a list of Worker PCS that would be updated once workers are started, or data is received: Columns: Worker Number, Worker IP, DateTime Request Started, DateTime Request ended, Result delivered (or alternatively saying “Running - waiting for Results..??), EC2 Status (“Requested start / Running / App running / Working / Requested Stop / Stopped ??" safe to exit now??) Whenever a worker has notified the result, a function has to be called; this function would then be used on our side to process the results as they come in (so independent of the GUI) public void NotifyResult (int worker, double Result, DateTime result_ended) { //We would put our code in here later. } **What is important** The reporting of Results back to the UserPC has to be as quick as possible. We therefore suggest to use a TCP socket (no .net serialization), and no encryption. Every millisecond counts. Going forward we want to optimize the application so that every microseconds counts as well. We would then do some tests on our side, to see if this approach makes sense or not. Depending on this, we would need some more fine-tuning going forward in the future (which is not part of this project). If you are familiar with TCP, UDP and modern webservers, you could perhaps assist us, to find ways how we could get the data from the webservers quicker than with our current implementation.