When i see what exactly is happening using the unix top command then i see that this process is using almost 85 to 90% memory and is varying even after the program is ended

Spreadsheet::ParseExcel is known to consume large amounts of memory for megabyte sized Excel files. See also Reducing the memory usage of Spreadsheet::ParseExcel.

But if i parse them individually then there is no problem

If your program can handle the files individually but not together then I'd guess that you are creating a new Spreadsheet::ParseExcel object for each file which, given the standard memory usage, is inefficient. If this is the case then try reuse the same object reference for each new file.

I have one more query is there a way to split a single excel file into multiple ones using some perl module or some thing.

Apart from Spreadsheet::ParseExcel the other main way would be to use Win32::OLE on Windows (or possibly on Linux).

Or can somebody suggest a way to convert a excel file into CSV(comma seperated values) format.

The OS spreadsheet application Gnumeric contains a command line utility that allows you to convert spreadsheets from one format to another:

    $ ssconvert simple.xls simple.csv
[download]

--
John.