Hi Animesh,
No answers, just questions
What benefit do you perceive in having a server running that (potentially) does not have a valid data file?
What happens when the client connects and the server then discovers that the file is corrupt? Do you tell the user? Or just kill the connection? Or try and fake it?
If you do tell the client that the data file is corrupt, what can they do about it?
If you were to validate the data file at the time the server is started, what can the person starting the server do about it? Is this more than what the client can do about it?
And to put this in real-life terms: If the production database doesn't work, you are likely to get paged to fix it. Would you rather get a single page from the person working on the server itself, or would you rather get 100 pages from clients all over the globe who can only tell you that the server is not working?
Regards, Andrew