|
North American Network Operators Group Date Prev | Date Next | Date Index | Thread Index | Author Index | Historical Re: SONET Interconnect (was RE: MCI)
Steve Steinberg <[email protected]> wrote: >I was talking with David Tennenhouse yesterday about why much of the >Internet community hates ATM and he said something that seems very >accurate: "Everyone worries about overhead in someone elses layer." Heh. He's telling that to carrier ISP people who are generally quite involved in level-2 transport issues? 30% of bandwidth is something to think of, when it costs more than all your equipment. >It's true: just think of how much TCP/IP overhead we put up with that >could be compressed if it was really important. Have you ever asked why IP compressors were't a smashing success? Answer -- because most of bulk traffic on Internet is compressed by hosts (be it GIF or JPEG, or good ol LZW compress). You can't really compress much of headers necessary for end-to-end reliability without per-hop header reconstruction (that is done on very slow links by PPP and C-SLIP, you can do that on T-1s with difficulty, but forget about in on T-3s). >(Not to mention HTTP overhead...) HTTP is being fixed. ATM is beyond repair. >Perhaps more importantly, stepping up fiber bandwidth is a >lot easier than improving router speeds. So why does a 15% loss due to >ATM really matter? It's 30%, for starters. There's a lot of short ACKs and keystrokes. Then, if "stepping up fiber bandwidth" isn't really a big deal why do we need ATM's QoS stuff? Just throw in more bandwidth! >Tennenhouse, btw, is proof that ATM did not come out of the telecom >community alone as people like to believe. You can write a FORTRAN program in any language. You can be a bellhead without being employed by a telco (and vice versa). >It's no suprise that they went with something like ATM that's good for >switch efficiency but bad for transmission line efficiency. That's a hogwash. ATM is bad for switching efficiency. That's obvious to anybody who ever heard about relation between size of memory footprint and speed, and can compare number of forwarding decisions per byte of user payload in IP and ATM. >Fortunately, that was probably the right tradeoff to make. That's apparently why there's no global ATM network on horizon but you can find Internet in Ust'-Kamenodyrsk. --vadim
|