ecomm_guru
asked on
Bandwidth Consumption
Can anyone let me know, what is the average bandwidth consumption taken per user for....
A PC---- I assumed 20kbps per user
A Laptop----- I assumed 20kbps per user
A Windows terminal on Citrix ----- I assumed 30kbps per user
Extra Bandwidth for Printing and Barcoding --- I assumed 20kbps per site if site has more than 30 users, 15kbps if between 10 and 20 users, 10kbps for 5 to 10 users and 5kbps for 1 to 5 users.
Could you please justify for me, if possible.
Thank you.
A PC---- I assumed 20kbps per user
A Laptop----- I assumed 20kbps per user
A Windows terminal on Citrix ----- I assumed 30kbps per user
Extra Bandwidth for Printing and Barcoding --- I assumed 20kbps per site if site has more than 30 users, 15kbps if between 10 and 20 users, 10kbps for 5 to 10 users and 5kbps for 1 to 5 users.
Could you please justify for me, if possible.
Thank you.
you can use NetMedic to measure the bandwidth used by a host
These numbers really depend on who's using the equipment. Applications, protocols, and such.
If you put me on one of those PCs, I'm going to throw your numbers way off. (Techies love to download lots of big files and such)
Better to get a little more detail and track the apps that people are using on the network and track that usage.
Tim
BrainBench MVP for TCP/IP Administration
If you put me on one of those PCs, I'm going to throw your numbers way off. (Techies love to download lots of big files and such)
Better to get a little more detail and track the apps that people are using on the network and track that usage.
Tim
BrainBench MVP for TCP/IP Administration
ASKER CERTIFIED SOLUTION
membership
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
ditto
No such thing as average, except for maybe 10 - 20% of users.
Some have to move files. Some merely hit little webpages. Some run applications that are more specific to company. Others are real heavy on eMail.
One of the heaviest users is the network traffic manager maintainers with their pings and snmp packets etc, used to be 10-20% bandwidth just to monitor network characteristics. Now they do more, and roll in A/V, IDS etc.
to estest - you may not realize how much you won't do. If downloading files of personal interest. I ran analyzer for awhile to ID rogue servers (developmental area, detecting new builds misconfigured to do routing and such) and found -- people running real-time games, bygosh. But bandwidth barely above casual user.
Main bandwidth users over time were corpos with their eMail. They fall behind the automagics - the non-human transmitters who broadcast how much they know (continuously) and the (unauthorized) backups. On occasion, equipment misconfigured, perhaps moved (continuously asking where RU?).
If this is about censoring people, I suggest you find area (project) where you can be more productive. There ever seems to be some VIP who knows that rules have to be bent for their special needs. And anyone who has ever had a 'problem' on wire, just may have had lunch with such VIP. In short, trying to maintain a useful rule-set is likely to become burdensome and less informative.
While there are things you can do to isolate things like running a tape backup across the network, there seems little that can be done over long term. Most people miss the point that devices like routers can send their own packets every second all day long, while mere humans take breaks. Or, take time to read or install what has been downloaded.
One thing you can manage a little, is the new-wave of live video and live audio. We have (fairly well) stopped that (so dunno what kind of percent that could be, I just hear it is very high on bandwidth, possibly comparable to running tape backups for all servers).
----
You seem to be comparing platforms with insufficient information available. I ditto all comments above re: "depends...".
For example, there is NO difference between laptop, notebook, and desktop. UNLESS... one is used to run a remote connection to other, such as using pcAnyWhere, et al. In which case, which end are you counting, remote or host??? And, what about the broadcasts they do that are independent of access.
For printing, does that not have to do with load? Is the shop a mailer that is constantly printing all day long?
For Citrix types, the proportion can go either way. While it used to be that people would run word processing off the network, causing traffic for the .exe's etc., nowadays with cheap-o HD's, more seem to have all apps on local HD, no bandwidth used until it is time to share new document or speadsheet or presentation with others. With less use for server-based info, their bandwidth would be less than Citrix type. But that makes it apples/oranges compare. What if users had to always run apps off server(s)? This is important for admin control, to maintain common platform (and maintaining upgrades, patches, consistency of templates, etc.).
Comparing bandwidth based on qty of users is also too relative. There is no such animal. IMO.
For real world, you need to sample the real wire, the real usage, before coming up with any specs. Two things will come to light very quickly, I venture, and both have to do with the clock on the wall.
(1) There will be a prime time vs non-prime period. This will be again up to applications and usage. But it exists. In olden times this was a breaking of 24 hour day into two equal periods of 12 hours each. Assumption held that daytime had more bandwidth requirements than night-time or weekends. Does not hold up so well if servers backed up at night and people at home doing remote access after returning from work or if 24 hour shop handling customer support calls. I say the period remains definable, but, it cannot be simply 'assumed' any more, you have to actually look at the traffic to find out what your case is.
(2) There remains the 'busy hour'. There is for any reasonable SLA a requirement to handle well the busiest time of day. Assumption is that if the high-traffic period can be handled well, that the rest of the day/week will follow. My prediction: this will be the TOD (time of day) when the majority of users are about to go to lunch.
No such thing as average, except for maybe 10 - 20% of users.
Some have to move files. Some merely hit little webpages. Some run applications that are more specific to company. Others are real heavy on eMail.
One of the heaviest users is the network traffic manager maintainers with their pings and snmp packets etc, used to be 10-20% bandwidth just to monitor network characteristics. Now they do more, and roll in A/V, IDS etc.
to estest - you may not realize how much you won't do. If downloading files of personal interest. I ran analyzer for awhile to ID rogue servers (developmental area, detecting new builds misconfigured to do routing and such) and found -- people running real-time games, bygosh. But bandwidth barely above casual user.
Main bandwidth users over time were corpos with their eMail. They fall behind the automagics - the non-human transmitters who broadcast how much they know (continuously) and the (unauthorized) backups. On occasion, equipment misconfigured, perhaps moved (continuously asking where RU?).
If this is about censoring people, I suggest you find area (project) where you can be more productive. There ever seems to be some VIP who knows that rules have to be bent for their special needs. And anyone who has ever had a 'problem' on wire, just may have had lunch with such VIP. In short, trying to maintain a useful rule-set is likely to become burdensome and less informative.
While there are things you can do to isolate things like running a tape backup across the network, there seems little that can be done over long term. Most people miss the point that devices like routers can send their own packets every second all day long, while mere humans take breaks. Or, take time to read or install what has been downloaded.
One thing you can manage a little, is the new-wave of live video and live audio. We have (fairly well) stopped that (so dunno what kind of percent that could be, I just hear it is very high on bandwidth, possibly comparable to running tape backups for all servers).
----
You seem to be comparing platforms with insufficient information available. I ditto all comments above re: "depends...".
For example, there is NO difference between laptop, notebook, and desktop. UNLESS... one is used to run a remote connection to other, such as using pcAnyWhere, et al. In which case, which end are you counting, remote or host??? And, what about the broadcasts they do that are independent of access.
For printing, does that not have to do with load? Is the shop a mailer that is constantly printing all day long?
For Citrix types, the proportion can go either way. While it used to be that people would run word processing off the network, causing traffic for the .exe's etc., nowadays with cheap-o HD's, more seem to have all apps on local HD, no bandwidth used until it is time to share new document or speadsheet or presentation with others. With less use for server-based info, their bandwidth would be less than Citrix type. But that makes it apples/oranges compare. What if users had to always run apps off server(s)? This is important for admin control, to maintain common platform (and maintaining upgrades, patches, consistency of templates, etc.).
Comparing bandwidth based on qty of users is also too relative. There is no such animal. IMO.
For real world, you need to sample the real wire, the real usage, before coming up with any specs. Two things will come to light very quickly, I venture, and both have to do with the clock on the wall.
(1) There will be a prime time vs non-prime period. This will be again up to applications and usage. But it exists. In olden times this was a breaking of 24 hour day into two equal periods of 12 hours each. Assumption held that daytime had more bandwidth requirements than night-time or weekends. Does not hold up so well if servers backed up at night and people at home doing remote access after returning from work or if 24 hour shop handling customer support calls. I say the period remains definable, but, it cannot be simply 'assumed' any more, you have to actually look at the traffic to find out what your case is.
(2) There remains the 'busy hour'. There is for any reasonable SLA a requirement to handle well the busiest time of day. Assumption is that if the high-traffic period can be handled well, that the rest of the day/week will follow. My prediction: this will be the TOD (time of day) when the majority of users are about to go to lunch.