I have been reading a whitepaper about video optimisation in 4.1 software (Accelerating Video Using Cisco Wide Area Application
Services), and would like some more information about quote from this paper:
"Typically, Windows Media Player (WMP) will initially request a User Datagram Protocol (UDP) connection to the video server in the data center. When the Cisco WAAS video application optimizer identifies this request as a request to join a Windows Media live event, it will deny this request for a UDP connection. WMP, as part of its default behavior, will resend the request, this time asking for a TCP connection."
I am interested to know whether this behaviour is only applicable in an inline deployment, or if there is some way that this could work for an off-path deployment using WCCP? I would think that in a WCCP setup, this original UDP setup request would never reach the WAAS appliance and therefore the WMP protocol rollover will not take place thus resulting in the connection receiving no benefit from the basic layer 4 optimisation, let alone the additionally licensed Video AO. Is this correct? Or is there some way to get the benefits of optimisation for this type of connection?
Location of whitepaper: http://www.cisco.com/en/US/prod/collateral/contnetw/ps5680/ps6870/white_paper_c11-499855.pdf
The initial connection setup between the Windows Media Player and the streaming server is done over a TCP control connection. This is the connection that is intercepted by the WAAS device and interperted. If the request is to set up a UDP streaming connection, WAAS the Video AO will deny the request. The WMP will then request a TCP connection which will be accepted and passed on to the streaming server. Because the initial control connection is a TCP connection, it doesn't matter whether the interception methed is inline or WCCP.