Article View: chi.internet
Article #46847Re: timings - real world thruput
From: Karl Denninger
Date: Wed, 05 Jul 1995 00:00
Date: Wed, 05 Jul 1995 00:00
76 lines
3487 bytes
3487 bytes
In article <3tdln0$8fe@news1.mcs.net>, Phil N. Schuman <pschuman@interserv.com> wrote: >I've seen various comments concerning the speed of ISDN when used >with different types of DCE. > >However - I think it would be most useful to see a set of real world figures >for a GIF or ZIP FTP, and/or a WEB page transfer for the following: Ok :-) 25kbps >28.8 modem 50kbps >ISDN TA - single B channel - V.120 rate adaption 55kbps >ISDN TA - single B channel - async/sync PPP connection 63kbps >ISDN ethernet router - single B channel 126kbps >ISDN ethernet router - both B channels These are for compressed data. Uncompressed data gives much higher rates for the last two options if you're using a compatible device. For example, an FTP of a postscript document via a 2B ISDN connection on a P50 here gives nearly 260kbps in real throughput. Now THAT is fast! >of course - the timings would have to be on the same file, and tried a couple of times >to ensure some consistent stats. >But, I would like to see the bits/sec and the overall >transfer time for the file transfer. See above. >any other testing conditions, discussion, or stats would be greatly appreciated. > >Phil - There are two issues to throughput: Latency Raw bit rate The first affects interactive connections severely. TAs in general are unable to get below 100ms on a "ping" response. This is due to the packetization that must be done with either V.120 or Sync/Async conversion to get the data from one end to the other. This is, by the way, pretty close to what a V.34 modems can give you in terms of interactive responsiveness. Not bad, but not great either. Raw bit rate varies less. TAs generally will only run an effective 57.6kbps DCE rate, although we have seen some which will clock at 115.2kbps. The problem then becomes having a serial port which will accept data at that speed without dropping it. Very few PCs can handle this without an add-on card like the Hayes ESP board. Ping times look like this: TA V.120 >> TA V.120 ~ 150ms TA Sync/Async >> Ascend MAX ~ 100ms Ethernet 1B >> Ascend MAX ~ 30ms (!) Ethernet 2B >> Ascend Max ~ 23ms (!) Note that the 1B return speeds are faster than a dedicated 56kbps circuit, which runs about 35 - 40 ms. As such it is quite impressive. 2B connections are 128kbps in real throughput. All these numbers were taken from my testing at MCSNet between my home environment and the office. The TA figures are a couple of months old -- the others are current (within the last two to three weeks). -- -- Karl Denninger (karl@MCS.Net)| MCSNet - The Finest Internet Connectivity Modem: [+1 312 248-0900] | (shell, PPP, SLIP, leased) in Chicagoland Voice: [+1 312 248-8649] | 7 Chicagoland POPs, ISDN, 28.8, much more Fax: [+1 312 248-9865] | Email to "info@mcs.net" WWW: http://www.mcs.net ISDN - Get it here TODAY! | Home of Chicago's only FULL AP Clarinet feed!
Technical Details
Message-ID:
<3te83b$o7q@Venus.mcs.com>Path:
arc1-ca.pugleaf.net!unknown!not-for-mailReferences:
<3tdln0$8fe@news1.mcs.net>Show Full Headers
Complete Article Headers
X-Google-Language: ENGLISH,ASCII-7-bit X-Google-Thread: 10d5fa,32e71bd291fee183 X-Google-Attributes: gid10d5fa,public X-Google-Thread: 13237b,32e71bd291fee183 X-Google-Attributes: gid13237b,public From: karl@MCS.COM (Karl Denninger) Subject: Re: timings - real world thruput Date: 1995/07/05 Message-ID: <3te83b$o7q@Venus.mcs.com> X-Deja-AN: 105924060 references: <3tdln0$8fe@news1.mcs.net> organization: MCSNet Ops, Chicago, IL newsgroups: chi.internet,comp.dcomp.isdn