Silicon Investor (SI) -- The First Internet Community

STOCKTALK

We've detected that you're using an ad content blocking browser plug-in or feature. Ads provide a critical source of revenue to the continued operation of Silicon Investor. We ask that you disable ad blocking while on Silicon Investor in the best interests of our community. If you are not using an ad blocker but are still receiving this message, make sure your browser's tracking protection is set to the 'standard' level.

Non-Tech : Quote.com QCharts -- Ignore unavailable to you. Want to Upgrade?

To: Sean W. Smith who wrote (1946)	5/3/1999 9:36:00 PM
From: Dan Clark	Read Replies (2) \| Respond to of 17977

Judy, actually there is one more format: =NASDAQ:CSCO =NYSE:IBM =NASDAQ:DELL etc... If you use this format, QFeed does not have to look up the symbols. It's a little faster. Regards, Dan.

To: Sean W. Smith who wrote (1946)	5/3/1999 11:04:00 PM
From: Sean W. Smith	Read Replies (1) \| Respond to of 17977

All, WARNING: Long post on understanding Round Trip Propogation Delays.... From: "Sean W. Smith" <sean_smith@mindspring.com> Liam and All, long reply below.... > > From: "Liam" <liam@fix.net> > > Ping Plotter has an option of how many "Samples to Include" . Type a zero > in there and it will change to "ALL". > > Now, when you let the thing run it will collect the entire day's > timeouts in > the "Err" column. If you don't see "err" just click on the column label > "PL%" and it will pop up next door. .... unless you have an old version. > > These errors due to timeouts will appear next to the router or server that > created them. You may be surprised how many are due to Quote.com servers > and the routers that feed the server in your connection. I was. > > A timeout means data did not get handled and was disposed of when too much > time expired it. It might have other meanings, but I'm a beginner at this > out of necessity. > > A QCharts chart might not be written completely without the complete data > package to do the job. A timeout might lose an essential chart > number and we > see it as QCharts slowing down. Exactly. Those packets must be re-trasmitted successfully to complete the data sequence. Some protocols can receive packets our of order and re-assemble them. Others used whats called a goback to N algorithm where all packets starting from the one dropped must be retrasmmited... > > Now, please, someone (hello, Sean) correct these assumptions or > add to them > so we know the score. OK, I add in some. I told you I'd post explaining the network delays and why the #'s vary greately and what can be implied from them.... Some of the descriptions Below are taken from "Data Communicaions,Computer Networks and OSI" by Fred Halsall which is a undergrduate text often used with introduction to computer networking. "Associated with any form of transmission medium there is a short but finite time delay for an electrical ( or optical signal) to propogate (travel) from one side of the medium to the other. This is a know as the transmissions propogation delay of the medium. At best Electrical signals propogate through free space at the speed of light (3x10^8 Meters/Second). In a physical transmission medium such as twisted pair or coxial cable, however, the speed of the propogation is a fraction of this figure. Typically 2x10^8 m/s or 0.5 x 10^-8 s/m." "As will be seen in subsequent chapters, data are normally transmitted in blocks (also known as frames, packets, or cells) of bits an on recepit of a block, an acknowledgement of the correct (or otherwise ) recepeit is returned to the sender. An important parameter therefore is the round trip propogation delay associated with the link. that is the time delay between the first bit of a block being transmitted by the sender and the last bit of its associated acknowledgement being received." <<End Quotes>> In addition there are delays introduced by the "protocol stack" on each end point that also factor. One must also consider that every layering or encapsulation of a protocol involves a loss in effiency because of overhead for various functions such as header, source address, destination address, Pointers to High Level Protocols (Saps, Ports), and CRC (Cyclic Rundancy Check) or FEC (Forward Error Correction). finally one must considering congestion or traffic shaping which add inherent queueing delay. so lets look at the formulas and see what they tell us.... I will make the following assumptions in my calculation. Protocol Stack Delay = 100us or 100x10^-6 seconds Propogation Speed of Electrical Signal through the medium = 2 x 10^8 meters/second Traffic Shaping, Queuing or Congestion Delay will be considered to be 0. I will factor in efficieny but assuming 10 bits/byte. This assumes that roughly 25% of all gross bits are used for overhead and not for actual data delivery. (This # in modern systems is typically higher. Quality of Service and desire for high levels of abstraction and portability via encapsulation are driving this higher). So lets start with 28.8K modem and Analog Phone Line. I will use my house as an example.... I am ~16,000 Meters from a Central Office Switch and lets use a 64 Byte packet as an example size. Trtpd = (Tx + Tp + Tpstack) * 2 Tp = 16,000 * 0.5 x 10-8 = 80 * 10-6 Seconds Tx = (64 Bytes * 10 bits/byte) / 28,800 bits/second = 20 miliseconds tpstack = 100 us Therefore Trptd = ( 80 us + 20000 us + 100 us ) = 42,800 us or 42.8 Milliseconds... This is purely point to point and would be considered one hop in IP. You would have repeat this calculation for each hop to get a total #. Now look at the same equation and substitute DS3 line of the same distance line @ ~45M bits/second. Tp = 16,000 * 0.5 x 10-8 = 80 * 10-6 Seconds Tx = (64 Bytes * 10 bits/byte) / 45,000,000 bits/second = 14.2us tpstack = 100 us Therefore Trptd = ( 80 us + 14.2 us + 100 us ) * 2 = 384 us Now lets change the distance to 5000 Kilometers Tp = 5,000,000 * 0.5 x 10-8 = 25 Miliseconds Tx = (64 Bytes * 10 bits/byte) / 45,000,000 bits/second = 14.2us tpstack = 100 us Therefore Trptd = ( 25000 us + 14.2 us + 100 us ) * 2 50,228.4 us or 50 miliseconds. Big Difference. Its easy to see from this example why a cable modem is much faster than traditional modem and why a connection from here to california adds significant delay. There are many other factors that effect this. Light through single mode fiber is faster than electrons through copper. Queueing Delay and delays caused my traffic shaping algorithms used in ATM networks are common. These make up most of the difference you see my theoretical #'s used here and ping plotter. The discussion of Queueing theory is beyond the scope of my time buget for this post. The above cited text is a good introduction to the subject. Its focused more on the theory and math behind networking than talking about any specific network protocol. For more info on TCP/IP and associated protocols: Most books on Windows NT will have some disucssion of IP. TCP/IP Illustrated is probably the most comprehensive book on IP that is readable, IMO. The TCP/IP unofficial bibles by Comer volumes I, II , and III will bore all but the most diligent student and I don't recommend except for serious developers. questions, comments, corrections all welcome.... Sean ------------------------------------------------------------------------ Did you know that onelist.com More than 3.5 million people are using ONElist? ------------------------------------------------------------------------ To unsubscribe from this list, go to the ONElist web site, at www.onelist.com, and select the User Center link from the menu bar on the left. This menu will also let you change your subscription between digest and normal mode. ONElist has created a URL that people can use to view the archives of your list. The URL is: onelist.com