The problem is that the unmetered option is not sustainable at the prices that are currently being charged. The ISP's (telco/cableco, whatever) have based their unlimited pricing structure on people only using 10% of their capacity at any point in time (that's an example, I have no idea what the telco projections are/were).
Now that people are using more capacity than what they thought they are left with two options:
1. Limit what people can use 2. Increase prices to allow for the increased usage.
I would contend that unmetered access is not a sane business model unless you charge people for 100% of the capacity they have 100% of the time. This means that pricing would increase massively as people would be forced to pay for the speed of their access.
Do you want to pay something like: 1mbps = $200/mon 2mbps = $350/mon 10mbps = $1500/mon (etc)
This is unlimited pricing based on your access speed, which seems to be what you're advocating.
I've put one of our sales guys onto this slashdot story as a lead;)
We (the company I work for) actually develop some other software for QH, so you never know, something might comes of it...
(sorry I misread that you were looking for scheduling for doctors, which as you say has different parameters again. I also think the people of my wife's ward have had it too good for too long with their personalised rosters.)
My wife is the same (works as a nurse in an Australian hospital) and asked me a similar question about 12-18 months ago. At the time she was sick of always getting a bum roster because she wasn't buddies with the people who worked out the roster. I originally thought that it would be easy, just put in a bunch of variables and it would spit out an optimal roster. Then I actually started thinking about it and there are a lot of variables that are hard to actually quantify into hard values to put into some sort of program.
Some of the inputs would include:
* shift lengths - staff on this ward can work either 8,10 or 12 hour shifts * number of shifts - staff can be anywhere from 1 shift per fortnight up to 10 * time between shifts - union mandates that staff must have X number of hours between successive shifts * number of staff required during certain periods of the day - daytime needs more staff than nightime (when patients are sleeping)
I know what you're thinking, most of those variables you can put down into hard facts, true, but ones you can't are stuff like:
* each staff gets X red requests per roster and Y yellow requests. Red requests are days the staff would REALLY like to have off (eg. family members Birthday or similar). Yellow requests are days that staff would prefer to have off if given a choice. * some staff prefer to work weekends, some don't. This is based on personal preference of having time off on the weekend (to spend with family) or working on the weekend and getting a higher rate of pay. Same goes for night (eg. graveyard) shifts. * A program could spit out a roster that technically fits the criteria, but from a human perspective is less workable. Eg. working alternate morning then evening shifts all week. The better option would be to work ALL morning or ALL evening shifts. * Some staff don't get along with other staff. I think they need to grow up a bit, but this does indeed happen, some staff refuse to work with others they don't get along with. If they somehow get accidentally rostered on with someone they don't like, they take a sick day (which costs the hospital as they have to get temp staff in to cover).
These are just a few things that come to mind. In the ned I realised that it was probably a situation where a program wouldn't be able to do it as well as a person. The reason is simply because there were to many personal factors to take into account and that these factors would change from one ward to the next based on the people in each ward.
The whole IBM powerPC lineup (i/p/z) is now capable of what you have described for z/OS. Both iSeries & pSeries have had it for a few years, but with the p5 chip, IBM is REALLY starting to push it, as the p5 chip has some muscle to throw around and it really does become possible to use virtualisation to consolidate multiple servers on a single partitioned box.
Hardware & software (ie. using VMWare/Xen) virtualisation are not quite the same thing .
If you have 5 servers in a rack that are doing whatever they and only pushing an average 15% utilisation each, you can consolidate into a single physical box, partition it into 5 machines.
This saves you on: * rack space * power (electricity) * cost (only have to buy 1 server, not 5)
That's for a server envirnoment. As Apple is traditionally targeted as a desktop, then it would allow you to do the same thing. How may people do you know who have 2 or 3 computers at home, connected with a KVM switch ? You'd be able to have a single box running all your different OS.
For a "real" article on Internet traffic, you may want to have a look at this one: abstract Full article (PDF).
It basically says that the US Internet backbone traffic is doubling every year.
taken from the article: =============== Table 1.3. Traffic on Internet backbones in U.S.. For each year, shows estimated traffic in terabytes during December of that year.
I'd also like to point out to keep in mind that you are in the Phillipines, and as such, there is no guaranteeing the bandwidth available into your country. I say this as someone who lives in Australia, and am constantly bemoaning the lack of bandwidth, not between our Co & ISP, but between AU & the rest of the world.
I'd say from the figures you're giving, you should be very happy, I know I would be with those kind of speeds:)
As a consultant to companies running both Exchange servers & Domino servers, I would pick a Domino/Notes setup anyday over Exchange/Outlook.
Some reasons:
1. Choice of hardare - Domino will run on most popular platofrms (yes even liunx !). Exchange == stuck with NT.
2. Domino scales better.
3. Domino provides all of the functionality of Exchange (IE groupware stuff, calendars, meeting, etc)
4. Domino uses a _real_ Database as a backend, not just a glorified MS Access DB.
Q. So, if I was based with having to give groupware functionality to about 1000 mail users, what would I do ?
A. Buy a _single_ AS/400, run Domino natively on it, rollout Notes client, and live happily ever after:)
Tony.
(I would like to express that I work for a company with close ties to IBM, and hence my opinion may be biased, but I do honestly believe that Domino/Notes is a better solution.)
A few links for you to look at (Domino vs Exchange):
I've recently installed an 8-port MasterView KVM switch (model cs-128 I think), and it is working fine.
One thing you should watch if you have IBM PC's, is that IBM uses the 6th pin on the PS/2 style keyboard connector. This can cause problems with some mice/keyboard combinations.
The solution from the FAQ on the web site referenced in the manual is to clip the 6th pin off. I thought this a bit drastic (it may work ?), and instead tried a different mouse. Otherwise I was getting boot errors saying that there was no mouse connected, even though the switch supplies the correct signals to make the PC's think that it has a mouse connected. I ended up using an older stlye IBM mouse (circa 1997), and had no troubles.
Slashdot Mirror
The problem is that the unmetered option is not sustainable at the prices that are currently being charged. The ISP's (telco/cableco, whatever) have based their unlimited pricing structure on people only using 10% of their capacity at any point in time (that's an example, I have no idea what the telco projections are/were).
Now that people are using more capacity than what they thought they are left with two options:
1. Limit what people can use
2. Increase prices to allow for the increased usage.
I would contend that unmetered access is not a sane business model unless you charge people for 100% of the capacity they have 100% of the time. This means that pricing would increase massively as people would be forced to pay for the speed of their access.
Do you want to pay something like:
1mbps = $200/mon
2mbps = $350/mon
10mbps = $1500/mon
(etc)
This is unlimited pricing based on your access speed, which seems to be what you're advocating.
I've put one of our sales guys onto this slashdot story as a lead ;)
We (the company I work for) actually develop some other software for QH, so you never know, something might comes of it...
(sorry I misread that you were looking for scheduling for doctors, which as you say has different parameters again. I also think the people of my wife's ward have had it too good for too long with their personalised rosters.)
My wife is the same (works as a nurse in an Australian hospital) and asked me a similar question about 12-18 months ago. At the time she was sick of always getting a bum roster because she wasn't buddies with the people who worked out the roster. I originally thought that it would be easy, just put in a bunch of variables and it would spit out an optimal roster. Then I actually started thinking about it and there are a lot of variables that are hard to actually quantify into hard values to put into some sort of program.
Some of the inputs would include:
* shift lengths - staff on this ward can work either 8,10 or 12 hour shifts
* number of shifts - staff can be anywhere from 1 shift per fortnight up to 10
* time between shifts - union mandates that staff must have X number of hours between successive shifts
* number of staff required during certain periods of the day - daytime needs more staff than nightime (when patients are sleeping)
I know what you're thinking, most of those variables you can put down into hard facts, true, but ones you can't are stuff like:
* each staff gets X red requests per roster and Y yellow requests. Red requests are days the staff would REALLY like to have off (eg. family members Birthday or similar). Yellow requests are days that staff would prefer to have off if given a choice.
* some staff prefer to work weekends, some don't. This is based on personal preference of having time off on the weekend (to spend with family) or working on the weekend and getting a higher rate of pay. Same goes for night (eg. graveyard) shifts.
* A program could spit out a roster that technically fits the criteria, but from a human perspective is less workable. Eg. working alternate morning then evening shifts all week. The better option would be to work ALL morning or ALL evening shifts.
* Some staff don't get along with other staff. I think they need to grow up a bit, but this does indeed happen, some staff refuse to work with others they don't get along with. If they somehow get accidentally rostered on with someone they don't like, they take a sick day (which costs the hospital as they have to get temp staff in to cover).
These are just a few things that come to mind. In the ned I realised that it was probably a situation where a program wouldn't be able to do it as well as a person. The reason is simply because there were to many personal factors to take into account and that these factors would change from one ward to the next based on the people in each ward.
The whole IBM powerPC lineup (i/p/z) is now capable of what you have described for z/OS. Both iSeries & pSeries have had it for a few years, but with the p5 chip, IBM is REALLY starting to push it, as the p5 chip has some muscle to throw around and it really does become possible to use virtualisation to consolidate multiple servers on a single partitioned box.
Hardware & software (ie. using VMWare/Xen) virtualisation are not quite the same thing .
This isn't the first generation of powerPC chips with virtualisation features, it is the third (powerPC-4, powerPC-4+ now powerPC-5)
8 /ibm_shatte rrs_tpcc/
Obviously you haven't seen any of the articles around about the performance of the new range of IBM powerPC chips.
Go educate yourself:
http://www.theregister.co.uk/2004/11/1
one word: CONSOLIDATION
If you have 5 servers in a rack that are doing whatever they and only pushing an average 15% utilisation each, you can consolidate into a single physical box, partition it into 5 machines.
This saves you on:
* rack space
* power (electricity)
* cost (only have to buy 1 server, not 5)
That's for a server envirnoment. As Apple is traditionally targeted as a desktop, then it would allow you to do the same thing. How may people do you know who have 2 or 3 computers at home, connected with a KVM switch ? You'd be able to have a single box running all your different OS.
To download a full installation copy of IE, follow the instructions on this page:
http://www.broomeman.com/support/wsiedown.html
credit goes to my work colleague John who pointed this out (Hi John if you're reading !)
good cisco paper that explains the whole MTU, MSS, fragmentation business (includes pretty pictures :)
o logies_white_paper09186a00800d66f2.shtml
Also explains how this relates to GRE & IPSec tunnels not working.
http://www.cisco.com/en/US/tech/tk648/tk369/techn
For a "real" article on Internet traffic, you may want to have a look at this one:
abstract
Full article (PDF).
It basically says that the US Internet backbone traffic is doubling every year.
taken from the article:
===============
Table 1.3. Traffic on Internet backbones in U.S.. For each year, shows estimated traffic in terabytes during December of that year.
year TB/month
1990 1.0
1991 2.0
1992 4.4
1993 8.3
1994 16.3
1995 ?
1996 1,500
1997 2,500 - 4,000
1998 5,000 - 8,000
1999 10,000 - 16,000
2000 20,000 - 35,000
===============
but in the end, you'll believe what you want to believe anyway...
I'd also like to point out to keep in mind that you are in the Phillipines, and as such, there is no guaranteeing the bandwidth available into your country. I say this as someone who lives in Australia, and am constantly bemoaning the lack of bandwidth, not between our Co & ISP, but between AU & the rest of the world.
:)
I'd say from the figures you're giving, you should be very happy, I know I would be with those kind of speeds
As a consultant to companies running both Exchange servers & Domino servers, I would pick a Domino/Notes setup anyday over Exchange/Outlook.
:)
d ocs/8A92E42E89828D26852568650071E18E
D Lotus.htm
. pl?/99/19/e03-19.74.htm
p ?ArticleID=7932
Some reasons:
1. Choice of hardare - Domino will run on most popular platofrms (yes even liunx !). Exchange == stuck with NT.
2. Domino scales better.
3. Domino provides all of the functionality of Exchange (IE groupware stuff, calendars, meeting, etc)
4. Domino uses a _real_ Database as a backend, not just a glorified MS Access DB.
Q. So, if I was based with having to give groupware functionality to about 1000 mail users, what would I do ?
A. Buy a _single_ AS/400, run Domino natively on it, rollout Notes client, and live happily ever after
Tony.
(I would like to express that I work for a company with close ties to IBM, and hence my opinion may be biased, but I do honestly believe that Domino/Notes is a better solution.)
A few links for you to look at (Domino vs Exchange):
http://www.lotus.com/developers/itcentral.nsf/w
http://www.microsoft.com/exchange/productinfo/Z
http://www.infoworld.com/cgi-bin/displayArchive
http://www.crn.com/components/search/Article.as
I've recently installed an 8-port MasterView KVM switch (model cs-128 I think), and it is working fine.
One thing you should watch if you have IBM PC's, is that IBM uses the 6th pin on the PS/2 style keyboard connector. This can cause problems with some mice/keyboard combinations.
The solution from the FAQ on the web site referenced in the manual is to clip the 6th pin off. I thought this a bit drastic (it may work ?), and instead tried a different mouse. Otherwise I was getting boot errors saying that there was no mouse connected, even though the switch supplies the correct signals to make the PC's think that it has a mouse connected. I ended up using an older stlye IBM mouse (circa 1997), and had no troubles.
just my $0.02
- bte