Bandwidth Calculations
From Isenet
For a more general discussion of network optimization, see Internet & Network Bandwidth & Monitoring & QOS
Software Tools
To measure your Internet connectivity speed, you can use:
- http://www.dslreports.com/stest
- http://wdc.speakeasy.net
- http://www.dnsstuff.com has a variety of tools
How Much is Enough?
One of the challenges as one moves to a 1:1 environment is determing how much bandwidth is enough. As part of that process I have developed a simple measure that I have found useful. Take the total available bandwidth and divide it by the number of end user CPUs in the network.
Using this technique, I surveyed several 1:1 independent schools in January of 2006. It appeared useful to separate them into two catagories managed and unmanaged as this had a substantial impact on relative satisfaction with the bandwidth available. Contributors are anonymous as I did not ask their consent for this use.
Managed (includes use of proxy server, packet shaper, or other management tool)
| State | Students | Grade levels | 1:1 Grades | Student Laptops | Faculty | Total CPU | Total Bandwidth | Bandwidth/CPU |
| CT | 820 | PK-12 | 7-12 | 500 | 120 | 750 | 1.5 MB | 2.0 K |
| Packet shaping is key, page rendering 2 to 3 seconds. Downloading during the day is "a joke." A separate circuit is used for tech support and video conferencing. | ||||||||
| CA | 295 | 9-12 | 9-12 | 295 | 40 | 410 | 3.0 MB | 7.3 K |
| Using NetEnforcer, blocking P2P and some games, bandwidth monitoring, feels they need more capacity. Increasing summer 2006. | ||||||||
| TN | 625 | 5-12 | 7-12 | 575 | 120 | 750 | 10.0 MB | 13.3 K |
| Using NetEnforcer & caching, blocking P2P, IM, some streaming, email throttled, audio files limited until 3:00 PM. Happy with performance. | ||||||||
| OH | 870 | K-12 | 5-12 | 550 | 130 | 850 | 11.5 MB | 13.5 K |
| Has Packeteer, but not shaping since last bandwidth increase. Gaming only permitted in one location. Happy with performance. | ||||||||
| TN | 787 | K-12 | 9-12 | 200 | 30 | 500 | 6.0 MB | 12 K |
| Use Packetsure for bandwith shaping and protocol blocking as well as URL blocking. Using Stratacache for caching services. Very happy with performance since moving from a T1 | ||||||||
Unmanaged (no proxy, packet shaping, or other active management. Filtering may be present.)
| State | Students | Grade levels | 1:1 Grades | Student Laptops | Faculty | Total CPU | Total Bandwidth | Bandwidth/CPU |
| CA | 1650 | K-12 | 9-12 | 675 | 180 | 1775 | 1.5 MB | 0.8 K |
| comment: | http, and ftp all unacceptably slow. Video conferencing on separate DSL circuit. Upgrade in progress. | |||||||
| TN | 740 | PK-12 | 7-12 | 320 | 75 | 600 | 3.0 MB | 5.0 K |
| comment: | No student use in unsupervised settings. Performance is ok. | |||||||
| FL | 1200 | PK-12 | 4-12 | 900 | 160 | 1100 | 14.0 MB | 12.7 K |
| No gaming or social networks during academic day. Happy with performance | ||||||||
| CT | 400 | 9-12 | 9-12 | 400 | 80 | 650 | 16.5 MB | 25.4 K |
| Caching is used. Some ad blocking. Happy with performance | ||||||||
Proposed rule of thumb based on this anecdotal evidence:
When bandwidth is managed, happiness is reached between 8 K and 13 K per CPU.
When unmanaged the required levels may be higher.
Other contributions to the pool of evidence would be welcome.
Dan Hudkins - The Harker School - San Jose, CA
