There is no single way to do this. The methods are generally dictated by the level of service you offer or your SLAs (Service Level Agreements).
Most ISPs work with an over subscription model. The assumption is similar to the way phone networks operate - you assume that only a portion of your customer base uses the network at any one time. If everyone starts to use the phone, the network will generally crash. Famous examples of this include the Bay Area in California during the Loma Prieta earthquake in 1989.
The key for you is to craft a model that reflects a) the service you desire to offer, b) the legal constraints of your agreements and state policies, and c) practical operating circumstances.
These are interactive. For example, you might include limitations in your service-restore-on-failure policy to reflect the difficulty of debugging a network at 2.00 am on Sunday morning. Most businesses will not care. Few private individuals will be using the net. Those that do, will probably accept such a limit if the other aspects of the service are good value.
Most ISPs run a ratio of signed members to backbone capability. Simple example: We want to offer a 1 Mbps service to users. We are provisioning an OC-3 connection (155 MBps). If we provision at 7:1, we can offer service to 1,085 customers before we have to upgrade our backbone with more capability.
Generally, this ration and consequences will be covered in our servce contract. Disclosing it is necessary to avoid complaints or legal action by customers during period of high traffic in the ISP network.
Customers usually fall into multiple categories, residential and business being the two most obvious. It is common for an ISP with a mix to create customer expectations in line with practical traffic. If you have a set of business customers who work 8-6 on five days a week, it is to your advantage to have lower residential traffic during this time. This may be reflected in lower service contract terms for residential customers during business hours. This is usually acceptable as most customers will also be at work. By contrast, the business user may get a lower service performance during non-business hours.
These variations show that any model must take each of the tiers into account. You can play all kinds of what-if scenarios to get a feel for the "right" mix for you.
Remember to allow for the common customer (e.g. the 5-day 8-6 office) and the more rigorous customer (7 x 24 operation or doing intense backups in the middle of the night). Having multiple pricing tiers and contracts usually cover these circumstances.
An older book that covers the general field and will give you food for thought would be "ISP Survival Guide - Strategies for Running a Competitive ISP" by Geoff Huston (Wiley).
If you find you do need more bandwidth to meet your planned/expected usage .... I suggest taking advantage of the free assistance available to find the right bandwidth solution for your specific application from DS3-Bandwidth.com.
Michael is the owner of FreedomFire Communications....including DS3-Bandwidth.com and Business-VoIP-Solution.com. Michael also authors Broadband Nation where you're always welcome to drop in and catch up on the latest BroadBand news, tips, insights, and ramblings for the masses.
Streamyx Coverage
0 Responses to “How To Predict Bandwidth Consumption In An ISP Network”
Leave a Reply