Services for ISPs

Almost every ISP grows through phases where they need to engineer aspects of their network which are new and novel to them. Since almost any mistakes in these areas can have drastic consequences, up to complete failure of the ISP and bankruptcy, and the right choices usually depend intimately on existing decisions that the ISP has made. This makes it hard to describe these tradeoffs.

With over 20 years of experience with The Internet and its precursors, I have seen an extremely wide selection of these interactions and can synthesize a solution that avoids not only the common pitfalls, but the esoteric ones as well.

Many times, the engineering staff of an ISP is fully committed to normal operations and planning for the future is put off until action is needed, at which point it's too late to put together a coordinated plan. This leads to a hurried design and deployment of these expansions, which often results in a design which works under normal circumstances, but has bad failure modes, and may be more fragile and fail more often. Additionally, networks that have grown through several of these rushed expansions, often become harder and harder to expand as the combinatoric interactions accumulate. Sometimes just measuring how close to the limit, and thus the time for action, a network is would help in avoiding these problems.

Specific services

These are descriptions of some of the areas that can be addressed, but this list is not exhaustive.

Routing

We can provide a full range of routing assistance. From simple router configurations, including BGP multi-homing, import/export rules, etc.; to complete routing architecture design and network planning services.

Robust multi-homing design

As ISPs get multiple connections to other providers for better connectivity, the complexity goes up enormously. It is difficult to get all the details right: proper fall back, effective filters, aggregation rules, etc. Frequently getting these wrong can make the network less reliable, which in the worst case can cause an ISP to fail.

The best way to get this right, is with the experience of having done it before. The best way to get the experience is by working with someone who knows what they're doing. I have lots of experience in setting up complex routing for global networks, and also in mentoring others in the intricacies of dynamic routing.

Routing architecture

There are many aspects that go into designing an overall network connectivity plan and routing architecture. You need to consider:

import/export rules: protecting yourself from bogus routes originating at customers, providers, peers, and other interconnects.
config management: controlling changes and generating redundant parts of configs to be compatible.
coordination with all other networks with whom you connect, via IRR, InterNIC, and other global databases; or through special arrangements.
automatic generation or verification of configs from IRR or other data sources.

A complete routing architecture takes time to design, time which your engineers may not have to spare right now. But, if you keep putting it off until there's time to do the whole project, you may never get there. This works for a while, as your engineers will have a common understanding, but it will constrain where the network can go.

I can provide an analysis of your existing network routing as well as your requirements and design an architecture to provide the maximum reliability, flexibility, and growth. By working with your engineering staff, I can produce a comprehensive network design / routing architecture document that can be used and maintained to keep future decisions consistent and to better plan the growth of the network. This will improve the productivity of your engineering staff, without requiring them to expend the time to document it.

Transition plans for "zero down-time" network changes

I have a lot of experience designing transition strategies that exhibit "zero down-time" to users. Renumber your web servers. Move mail servers from one location to another. Change DNS servers. Switch transit providers.

The Internet is global, there is no longer any "unused" time in which to take systems off the air for reorganization. With careful planning, however, it is usually possible to arrange a transition with no (or negligible) visible impact on users. I can design a spectrum of strategies for any transition with analysis of the costs and benefits of each option, and then guide you in choosing the tradeoff that is best for your customers.

Design of aggregation rules

Aggregation rules, both internal and for imported routes, are important to the growth and stability of any network. A well designed aggregation plan that maps to the network topology/connectivity design and the routing architecture of your network can help to make your network more reliable and can often lead to cost savings by allowing the use of lower cost routers in parts of the network.

Measurement and analysis

Measurement of current usage with trend analysis and projections of future capacity requirements can be a vital resource for network planning purposes. But at many ISPs, the engineering staff can be too busy to put in the time to evaluate and set up the various packages. I have experience setting up many of the freeware packages that can do this, and customizing them to a particular network.

Migration and expansion strategies for maximum flexibility

ISP engineers are often so overloaded with day-to-day issues that it becomes hard to make time to step back and look at the next migration/expansion step and design it fully. This often leads to plans which don't cover all contingencies, and can lead to the need for on-the-fly corrections, which is the source of many later problems. By carefully planning ahead, you can save much more in lowered maintenance costs.

Address space utilization plans

These days, address space allocation, renumbering, and CIDR aggregation are important to the ability of a network to grow. I have extensive experience in all these areas. Defining your address plan early and adjusting it as your network grows can help avoid excessive requirements to renumber and give you a better margin on additional allocations.

Since addresses are allocated according to RFC2050 in most cases, a good allocation strategy can help you reach the goals with more margin, giving you more lead time to get a new allocation in.

Staff training

I can provide training for your network operations and engineering staff, either train existing staff in new technologies being brought in, or to train new staff to keep from distracting your (probably overloaded) existing staff with this added task.

Especially useful is the combination of providing some engineering design from above and concluding with a complete, customized training series on the results. This will get your staff up to speed on the new setup as quickly as possible to allow them to maintain and extend it.

For more on the available training options, see the separate page with more details.

Consulting: [Overview] [Routing] [DNS] [Growth planning] [Training] [Product Deployment] [Content providers] [ISPs] [Enterprise nets]

MAP Net Eng.: [Main Page] [Consulting] [Software] [Hosting] [contact us] [Resume]

Customize the look of this site. There are several optional features you can turn on or off:

Left sidebar Table of contents: Presently off, turn it on
Top of page graphical tabs: Presently off, turn it on
Bottom of page textual navigation: Presently on, turn it off
Or, you could turn them all on

Web site designed, developed and hosted by MAP Network Engineering
MAP Network Engineering uses, supports and recommends Open Source software and open standards.

Use of the logos above does not imply endorsement by the respective organizations of MAP Network Engineering sites or services.
On the contrary it implies endorsement by MAP Network Engineering of those organizations or software.

And finally, a random note:
There are three kinds of people. Those who can count, and those who can't.