Thumbs up on this one; my entire path and chain of management of that path need to be equally fault tolerant - Awesome. On Thu, Sep 12, 2013 at 2:40 PM, Glen Wiley <glen.wiley@gmail.com> wrote:
Remember though that anycast only solves for availability in one layer of the system and it is not difficult to create a less available anycast presence if you do silly things with the way you manage your routes. A system is only as available as the least available layer in that system
For example, if you use an automated system that changes your route advertisements and that system encounters a defect that breaks your announcements then although a well built anycast footprint might acheive 99.999, a poorly implemented management system that is less available and creates an outage would reduce the number.
On Thu, Sep 12, 2013 at 4:25 PM, Phil Fagan <philfagan@gmail.com> wrote:
Its a good point about the anycast; 99.999% should be expected.
On Thu, Sep 12, 2013 at 2:14 PM, Beavis <pfunix@gmail.com> wrote:
I go with 99.999% given that you have a good number of DNS Servers (anycasted).
On Thu, Sep 12, 2013 at 9:03 PM, Phil Fagan <philfagan@gmail.com> wrote:
Everything else remaining equal...is there a standard or expectation for DNS reliability?
98% 99% 99.5% 99.9% 99.99% 99.999%
Measured in queries completed vs. queries lost.
Whats the consensus?
-- Phil Fagan Denver, CO 970-480-7618
-- () ascii ribbon campaign - against html e-mail /\ www.asciiribbon.org - against proprietary attachments
Disclaimer: http://goldmark.org/jeff/stupid-disclaimers/
-- Phil Fagan Denver, CO 970-480-7618
-- Glen Wiley KK4SFV
"A designer knows he has achieved perfection not when there is nothing left to add, but when there is nothing left to take away." - Antoine de Saint-Exupery
-- Phil Fagan Denver, CO 970-480-7618