RE: Software for Statistical Profiling [7:58277]
Thanks for the info - I was looking at using the ErlangC model due to the nature of computer networks. Currently, my modelling is based on assuming an application takes 2 seconds to complete a task if it had 32kbps of bandwidth reserved for it, I can use the Erlang model to calculate whether 95% of transactions can be completed within say 2.3 seconds (with the 0.3 seconds being queue delay) - but by changing the bandwidth I change my underlying assumption of 2 seconds. Ie, if I allocate 64kbps, then the application may only takes 1.2 seconds and my model changes accordingly. The traditional Erlang model applies well on a packet-by-packet basis (just like queues of people in the bank, etc), but I'm not sure of its applicability to a series of packets on a whole which make up a transaction. Regards Richard -Original Message- From: charles dunkirk [mailto:[EMAIL PROTECTED]] Sent: Saturday, 30 November 2002 12:38 AM To: [EMAIL PROTECTED] Subject: Re: Software for Statistical Profiling [7:58277] Try here for erlang calculator http://www.erlang.com/ There are 2 type B and C . B assumes blocked calls don't call back and C assumes they stay in queue. Chuck Dunkirk The Long and Winding Road wrote: >Richard, the software used for your telephone booth problem is called >an Ehrlang calculator. it seems that you could use an Ehrlang >calculator to do this as well. There are a number of web sites that >have Ehrlang calcs. A google search should reveal a bunch of them. > >A long time ago, in statistics class, we used to do something called >"monte carlo simulations" to figure out stuff like this also. I don't >remember much about the mechanics. Got a statistics professor on you >campus? > >Chuck > >-- >TANSTAAFL >"there ain't no such thing as a free lunch" > > > > >""Larkin, Richard"" wrote in message >[EMAIL PROTECTED]">news:[EMAIL PROTECTED]... > >>I recall in Uni that we used te Poisson distribution and some >>mathematical formulae to say that if we have x people arrive per hour >>at a phone booth, and the average phone call is y minutes, we would >>need z phone booths to ensure that 95% of the time, people don't have >>to wait (or only have to >> >wait > >>xx minutes). >> >>Transposing this to application budgeting, I have an application at a >> >remote > >>site which has a max of 5 concurrent users and the worst transaction >>they >> >do > >>will hog the 64kbps line for 30 seconds (if it is the only >>transaction). >> >>My question is without revising my lecture notes, what software would >>help me determine what bandwidth to allocate this application so that >>95% (or >>whatever) of the time the transaction can be completed in yy seconds? >> >>Is there any good software out there which would help me with this? >> >>Cheers >>Rik Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=58379&t=58277 -- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
Re: Software for Statistical Profiling [7:58277]
Try here for erlang calculator http://www.erlang.com/ There are 2 type B and C . B assumes blocked calls don't call back and C assumes they stay in queue. Chuck Dunkirk The Long and Winding Road wrote: >Richard, the software used for your telephone booth problem is called an >Ehrlang calculator. it seems that you could use an Ehrlang calculator to do >this as well. There are a number of web sites that have Ehrlang calcs. A >google search should reveal a bunch of them. > >A long time ago, in statistics class, we used to do something called "monte >carlo simulations" to figure out stuff like this also. I don't remember much >about the mechanics. Got a statistics professor on you campus? > >Chuck > >-- >TANSTAAFL >"there ain't no such thing as a free lunch" > > > > >""Larkin, Richard"" wrote in message >[EMAIL PROTECTED]">news:[EMAIL PROTECTED]... > >>I recall in Uni that we used te Poisson distribution and some mathematical >>formulae to say that if we have x people arrive per hour at a phone booth, >>and the average phone call is y minutes, we would need z phone booths to >>ensure that 95% of the time, people don't have to wait (or only have to >> >wait > >>xx minutes). >> >>Transposing this to application budgeting, I have an application at a >> >remote > >>site which has a max of 5 concurrent users and the worst transaction they >> >do > >>will hog the 64kbps line for 30 seconds (if it is the only transaction). >> >>My question is without revising my lecture notes, what software would help >>me determine what bandwidth to allocate this application so that 95% (or >>whatever) of the time the transaction can be completed in yy seconds? >> >>Is there any good software out there which would help me with this? >> >>Cheers >>Rik Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=58289&t=58277 -- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
Re: Software for Statistical Profiling [7:58277]
Richard, the software used for your telephone booth problem is called an Ehrlang calculator. it seems that you could use an Ehrlang calculator to do this as well. There are a number of web sites that have Ehrlang calcs. A google search should reveal a bunch of them. A long time ago, in statistics class, we used to do something called "monte carlo simulations" to figure out stuff like this also. I don't remember much about the mechanics. Got a statistics professor on you campus? Chuck -- TANSTAAFL "there ain't no such thing as a free lunch" ""Larkin, Richard"" wrote in message [EMAIL PROTECTED]">news:[EMAIL PROTECTED]... > I recall in Uni that we used te Poisson distribution and some mathematical > formulae to say that if we have x people arrive per hour at a phone booth, > and the average phone call is y minutes, we would need z phone booths to > ensure that 95% of the time, people don't have to wait (or only have to wait > xx minutes). > > Transposing this to application budgeting, I have an application at a remote > site which has a max of 5 concurrent users and the worst transaction they do > will hog the 64kbps line for 30 seconds (if it is the only transaction). > > My question is without revising my lecture notes, what software would help > me determine what bandwidth to allocate this application so that 95% (or > whatever) of the time the transaction can be completed in yy seconds? > > Is there any good software out there which would help me with this? > > Cheers > Rik Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=58286&t=58277 -- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]