This archive contains answers to questions sent to Unidata support through mid-2025. Note that the archive is no longer being updated. We provide the archive for reference; many of the answers presented here remain technically correct, even if somewhat outdated. For the most up-to-date information on the use of NSF Unidata software and data services, please consult the Software Documentation first.
John, > So going from .0175 to .03 seconds between packets in a TCP conversation is > enough to cause the latency we are seeing? According to <http://rtstats.unidata.ucar.edu/cgi-bin/rtstats/rtstats_summary_volume?uni14.unidata.ucar.edu>, about 6e9 bytes arrive every hour on the NIMAGE feed. From that, the rate of packet arrival can be approximated: (3600 seconds)/(6e9 bytes)(3000 bytes/packet) = 0.0018 seconds/packet If the time *between* packets is 0.03 seconds, then the network is too slow to keep up. > If you think that packet to packet (time between packets in a 388 > conversation) time has nothing to do with the latency we are seeing (and > wouldn't necessarily be seen while we are seeing the LDM latency), please > explain the physics to me because I'm seriously lost. I cannot for the life > of me understand how this latency could be occurring without me seeing the > packet to packet times increase dramatically. > I'm aware that networks can drop packets in a bad connection (where I'll see > Retransmits and Duplicates), block packets altogether (where there would be > RST ACKs most likely), or sequester/buffer/queue packets. The third I have > not seen in a packet capture and could only guess what it looks like (i.e. > many Retransmits with eventual new SYN requests as seen in a web server's > denial of service instances). If a packet-shaping system were discarding packets, then you wouldn't see any retransmissions because the TCP layers constantly recompute the retransmission-timeout (RTO) parameter based on the round-trip time (RTT). They do this to avoid retransmissions. A packet-shaper would, effectively, increase RTT and, hence, the RTO used by the TCP layer. > The packet capture you sent me had no traffic originating from mammatus. Did > you filter that out or was that a delay of 3 seconds? It looks filtered out. It was filtered out. What I sent only contained packets from us to you for the NIMAGE feed. Regards, Steve Emmerson Ticket Details =================== Ticket ID: GUQ-669331 Department: Support LDM Priority: Normal Status: Closed =================== NOTE: All email exchanges with Unidata User Support are recorded in the Unidata inquiry tracking system and then made publicly available through the web. If you do not want to have your interactions made available in this way, you must let us know in each email you send to us.