> On 22-Feb-15, at 08:47 , Roland Dobbins <rdobb...@arbor.net> wrote: > > > He should be looking for redundant active/active, instead. DNS, not an IP > address, should be used to reach each active instance of the service in > question. >
i think true active/active becomes a nirvana that we all would like to achieve from a technical perspective — however there are strong business implications that prevent achievement of said state. the discussion that should be happening is around the business continuity requirements of each of the services/applications in question. when the cost of an outage far exceeds even the fewest of moments of downtime, active/active should be the goal. however — through a detailed discussion from all of the different business units — it may be determined that active/active may be overkill for the service/application provided. things like “warm standby” or “cold standby” may be more tolerable from a capex standpoint — let alone the operational challenges in ensuring that all operations teams are ready to support a truly “active/active” design. while the ability to implement “things” provided by mpls technologies for transparent layer-2 connections (or through the use of $vendor technology) — coupled with storage federation and inbound routing correction through dns gslb or lisp is definitely doable — it may be determined that the cost is too great for everything — thus creating a tiering of redundancy services based on criticality to the business. otherwise — we’re all just being asked “how long is a piece of string?” q. _______________________________________________ cisco-nsp mailing list cisco-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/