Question

We have a webservice (so called “service with attachment”) where users can send files with their requests. Users can upload up to 256 MB of files. We have a working solution that parses the request and saves the files, but it’s slow. Sometimes when sending big attachments – server returns timeouts.

We read the request by bytes and looking for a boundary tags (--b5a8d09eeeb161be29def84633d6f6fc). If tag is found then parse the header and read the attachments file base64 encoded data. Also there’s a check for a newline characters.

In my opinion there’s a lot of cycles in this code and it can be done better and improve the performance.

Is there some great examples how is this done and maybe parsing a string (not a byte array) is faster solution? Or maybe there’s other approaches? What do you suggest to do to improve the performance?

Regards,
evilone

Answer 1

Ok, found the answer myself. Using Streamreader and ReadLine() function is about 40% faster than iterating over single bytes in byte[] array. If compression is added to requests too like Pareen suggested then it can be more faster.