Re: Network buffering

Tech-Archive recommends: Fix windows errors by optimizing your registry


"Jeremy Noring" <someone@xxxxxxxxxx> wrote in message
news:eP9fKRdfFHA.3912@xxxxxxxxxxxxxxxxxxxxxxx
>
> RTSPT is going to be built on top of TCP, so I'm assuming it's using TCP's
> built in retransmit mechanism. It wouldn't make sense to build a protocol
> on top of TCP and not use TCP to handle lost packets.

We're using RTSPT. I realize the difference between UDP and TCP, but those are
transport layer protocols. TCP can resend a lost packet up to a point--if the
packet has already been flushed from the network buffer then I believe you're
SOL. This is where application level protocols come into play. If TCP can't
recover, the application can by using its own reliable protocol. RTSPT is well
documented in RFC 2326 (http://www.ietf.org/rfc/rfc2326.txt). However, I just
read this excerpt which seems to imply that all retransmissions are handled via
TCP, not RTSP:

If a reliable transport protocol is used to carry RTSP, requests MUST
NOT be retransmitted; the RTSP application MUST instead rely on the
underlying transport to provide reliability.

If both the underlying reliable transport such as TCP and the RTSP
application retransmit requests, it is possible that each packet
loss results in two retransmissions. The receiver cannot typically
take advantage of the application-layer retransmission since the
transport stack will not deliver the application-layer
retransmission before the first attempt has reached the receiver.
If the packet loss is caused by congestion, multiple
retransmissions at different layers will exacerbate the congestion.

Another comment on the WMS buffer:

Windows Media Services adds a delay of approximately three seconds for network
buffering so that multiple live clients can pull content from the same buffer.
This buffer cannot be changed.


.

Relevant Pages

  • Re: load-balancing twin 10GbE?
    ... Ethernet with either 1500 byte MTU employing TSO - TCP Segmentation ... out-of-order segment a TCP receiver will generate an immediate ACK. ... if a sending TCP receives three duplicate ACKs ... retransmissions, which can result in smaller congestion windows which ...
    (comp.os.linux.networking)
  • possible solution to performance issues with FTP client on HP3000
    ... Maximum Time to Wait For Remote Response Initial Retransmission Interval ... performance in cases where single retransmissions are necessary and the "Maximum Retransmissions per Packet" of gives me the enough retransmissions to successfully keep a TCP connection going over a network that is not consistent in it's performance or quality characteristics! ... Common characteristics of "the internet" ISP's are inconsistent performance; burst of traffic can slow transfer of data buffered in switches; especially image, video and voice traffic - this can trigger TCP retransmissions. ...
    (comp.sys.hp.mpe)
  • retransmissions with bge driver?
    ... network performance, I see a lot of tcp retransmissions on both the ... the network cards are configured as ... Modifying the tcp kernel configuration as follows ...
    (freebsd-net)
  • Re: how to handle socket timeout?
    ... you want realtime behaviour, ie no retransmissions of old ... The server is a generic facility that implements both TCP and UDP- ... protocol, a logical thing would be to drop the response. ...
    (comp.unix.programmer)
  • Re: Coordinating TCP projects
    ... It is rapidly becoming clear that quite a few of us have Big Plans for the TCP implementation over the next 12-18 months. ... Jim and I recently discussed the idea of implementing autotuning of the TCP reassembly queue size based on analysis of some experimental work we've been doing. ... This means that if the TCP window grows to be more than 48 segments wide, and a packet is lost, the receiver will buffer the next 48 segments in the reassembly queue and subsequently drop all the remaining segments in the window because the reassembly buffer is full i.e. 1 packet loss in the network can equate to many packet losses at the receiver because of insufficient buffering. ... We observed that the socket receive buffer size provides a good indication of the expected number of bytes in flight for a connection, and can therefore serve as the figure to base the size of the reassembly queue on. ...
    (freebsd-arch)