A fun read. Sally Floyd, Craig Leres, Steve McCanne, Vern Paxson and of course Van Jacobsen, the LBNL Network Research Group, were heroes of mine earlier in my career. A terrible loss. Condolences to anyone here who know Floyd personally.
Van Jacobson explained it to us at Ford Aerospace around 1982 or so. Back then it was UUCP mail that was synching up. Mailers forwarded mail every 15 minutes or so, and they'd all synch up and try to call the same receiver at the same time. If you do "send mail, then wait 15 minutes", that happens, because the "send mail" part is variable time, longer if it's being held up by traffic. You have to start the next cycle N minutes after the start of the previous cycle, not the end, or you get strong synchronization.
https://www.icir.org/floyd/papers/sync_94.pdf
A fun read. Sally Floyd, Craig Leres, Steve McCanne, Vern Paxson and of course Van Jacobsen, the LBNL Network Research Group, were heroes of mine earlier in my career. A terrible loss. Condolences to anyone here who know Floyd personally.