I have a requirement to process a lot of data on a nightly basis. I have a network of semi-trusted PCs, an intranet - nothing external, mainly desktop computers, as well as some servers.
My initial thought was to use desktop timeouts as well as dedicated server resources as follows:
1) Linking data with encrypted 20-megabyte fragments (administrators do not want any data to be written to PC disks, they must remain in memory)
2) Distribute .net dll (s) containing the processing to be performed (often the processing changes, hence the distribution of the DLL)
3) Decrypt data
4) Execute dll methods
5) Return Results
The server will monitor the work units / tasks and after some time X, if the result is not obtained, check the status of the machine that should work on them and assign another machine, if necessary.
I would not classify the work as processor intensive work, however, a parallel element that would allow checking the entire database earlier is a real bonus.
I do not think that I need a machine-to-machine connection, just a client server. I am happy that the clients simply polled the server - I do not need anything complicated in the issue of notification of work.
My question is: are there any .NET structures that do such things? or is this approach completely flawed?
Jonno source
share