[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Granularity (was Re: [RRG] ALT + NERD is inelegant & inefficient, compared to APT or Ivip)
Brian E Carpenter wrote:
The idea is to make *less* work by splitting the problem space in two.
How does that apply in the case of a site that changes from single
homed to dual homed to triple homed and back again? I expect this
will be a very common occurrence.
If the instance of triple homed is really a question of switching which
second provider is being used,
then I think a suitable question to dig further along this is:
How important is it that while migrating, a site be able to use *all
three* links, if only the redundancy is critical?
And, how likely is it that the uninvolved link will fail during the
transition window for the other two links, as well as
one of the other two links failing simultaneously? Would that be any
worse than if both links failed, before or
after the transition event? Is having the third link available to do a
fast transition (or fast reversion) adequate
to meet the needs of a site which isn't so concerned with redundancy as
to operate their own EID infrastructure?
If we can generalize this to being, "It is common operational practice
to have an overlap period between
the provisioning of one service, and the cancellation of another
service, and a 'hot' cut-over between those
during the overlap duration", then I think it's okay to not support
temporary triple-homing.
Making the distinction between full-on triple-homing, and having three
links, two of which must always be
configured for redundancy, is IMHO a reasonable situation.
When there is one link that doesn't change, and two that are in the
hot-cut mode (one add, one drop),
the stability and redundancy is quite reasonable with a solution limited
to N=2. IMHO.
BTW - reminder about the N=2 distinction. I should point out, that the
"heavyweight" solution without
host granularity, is also available even in N=2. The compromise is, no
host granularity, and more administrative,
technical, financial, and operational overhead is required.
Brian Dickson
--
to unsubscribe send a message to rrg-request@psg.com with the
word 'unsubscribe' in a single line as the message text body.
archive: <http://psg.com/lists/rrg/> & ftp://psg.com/pub/lists/rrg
- References:
- [RRG] ALT + NERD is inelegant & inefficient, compared to APT or Ivip
- From: Robin Whittle <rw@firstpr.com.au>
- Re: [RRG] ALT + NERD is inelegant & inefficient, compared to APT or Ivip
- From: Dino Farinacci <dino@cisco.com>
- Re: [RRG] ALT + NERD is inelegant & inefficient, compared to APT or Ivip
- From: Robin Whittle <rw@firstpr.com.au>
- Re: [RRG] ALT + NERD is inelegant & inefficient, compared to APT or Ivip
- From: Eliot Lear <lear@cisco.com>
- Re: [RRG] ALT + NERD is inelegant & inefficient, compared to APT or Ivip
- From: Brian E Carpenter <brian.e.carpenter@gmail.com>
- Re: [RRG] ALT + NERD is inelegant & inefficient, compared to APT or Ivip
- From: Tony Li <tli@cisco.com>
- Granularity (was Re: [RRG] ALT + NERD is inelegant & inefficient, compared to APT or Ivip)
- From: Brian Dickson <briand@ca.afilias.info>
- Re: Granularity (was Re: [RRG] ALT + NERD is inelegant & inefficient, compared to APT or Ivip)
- From: Iljitsch van Beijnum <iljitsch@muada.com>
- Re: Granularity (was Re: [RRG] ALT + NERD is inelegant & inefficient, compared to APT or Ivip)
- From: Brian Dickson <briand@ca.afilias.info>
- Re: Granularity (was Re: [RRG] ALT + NERD is inelegant & inefficient, compared to APT or Ivip)
- From: Brian E Carpenter <brian.e.carpenter@gmail.com>