wiki:CtOperationsSurvey

Page to collect ideas for a survey of BIND 9 customers on what are the important operational issues that need to be addressed in migration to BIND10 and operation thereafter.

This seems too long for a survey as it is now. Which parts would best be culled and saved for a phone call. Maybe we can extend the contract customers for 1 or 2 days if they fill out the whole thing. :)


Intro [just my babble right now]

BIND 10 is a ground up restructure and rewrite of the BIND name server, a project that will run for a number of years. Most people have come to learn how to care and feed for BIND 9 within your operations organization. This is an opportunity for the development team to reach out the the operational community to find out what works well, what doesn't work well and what priorities should be set for the operational requirements of the new software.

Your input is very important to this process. If you would take a few moments of your time answering information about how your site does it's work, it will help us greatly.

Would you be willing to talk by phone with someone from ISC about BIND operational support?

BIND operation

First are some general questions about your DNS environment. These questions are looking for rough estimates.

Which type of organization would you say best fits you and your DNS needs

large corporation governmental large organization (NGO, University, etc.) small/medium organization ISP hosting/ASP web company registry other

How many name servers do you have today (if one machine provides both, count it in each)

authoritative recursive

number expected in 2 years

authoritative recursive

how many total resource records do you offer in zones you are master for?

today in 2 years

How many zones do you handle

master today slave only today master in 2 years slave only in 2 years

what drives naming decisions in your DNS (answer each 0-5 with 5 being most important)

branding/marketing ease of navigation (user predictability) internal user expectation external user expectation

DNS is a critical resource. if you were to take a 1 hour major DNS outage what would be your estimate of

how much money the company would lose from this outage what is the percentage likelihood that someone will be explaining to a VP/director why this will never happen again

What about a extra 2 second delay on all queries for 4 hours

how much money the company would lose from this slowdown what is the percentage likelihood that someone will be explaining to a VP/director why this will never happen again

The next set of questions are a 0-5 indication of importance to your organization. 0 means you never use this and 5 means you could not operate your DNS environment without this function. Each question is asked twice, once for where you are today and a second time for where you think you will be in 2 years.

simple logging and statistics complex logging complex statistics database backed zones instead of text files Dynamic DNS zone transfers by non-DNS methods (not using AXFR/IXFR/Notify) Anycast DNS service (same ip address multiple places in the network) Views (a single DNS server that offers different data for different clients) DNS based load distributor (Akamai, Speedera, Cisco DD, etc.) integration of larger operational process data into DNS

Now for some open ended questions:

What makes your site unique regarding DNS

What operational aspects of BIND 9 are the most difficult to deal with

What changes in BIND would most improve your operation of DNS servers

How could BIND 10 better integrate with other operational data sources

Do you have any thoughts about when and why you would consider moving to BIND 10

What features would be absolute requirements that would prevent migration if they weren't present

Last modified 7 years ago Last modified on Oct 28, 2010, 9:28:47 PM