While walking our dog this morning, I thought of a great example as to why this is a BAD idea.
- Let assume that you DNS team make a change and goes for lunch.
- You have 15000 PCs at Site A
- You have both HW and SW inventory enable, each will send a minimum of 5 MB each to the site server
- Assume that SW inventory take the full 4 hours to run
- 68 minutes for the DNS team to fix the problem
- CM07/CM12 client will automatically fail over to Site B after missing 1 attempt at contacting the CM site server.
- CM07 PC can resolve the Site B CM07 site server.
Based on the above, this will mean ALL clients from Site A will fail over to Site B, It will also mean that all client will be FORCE to generate a full inventory for everything (HW, SW, DCM. HB, SU, etc).
(15000 * 5 MB) for HW Inv + (15000 * 5 MB) for SW Inv + (15000 * ?? KB) for each DCM rule + (15000 * 1KB) for Heartbeat Inv
= minimum of 150015 MB or 150 GB if you fail it back to the old site once the DNS is fixed you will end up with yet another 150GB of bandwidth consumed and all of this because the DNS team make a mistake.
The question about HA is:
- What feature of CM07/CM12 does you requite up and for why?
- What does downtime really mean?
- What is affected by the downtime?
- How much down time can you afford? 1 hours hour, 1 Day, 1 week???
- Do you have other options to temporally solve you needed during a outage? (1 hours, 1 day, 1 week?)
- Do you have a plan to migrate client in the event of a DR or Emergency?
- Can you re-build the site in the allowable downtime?
Without answer these question, you cant design an HA / DR solution. I have yet to find a client where they cant live without CM07/CM!2 for at least 48 hours and most agree that 1 week is acceptable! As such they can easily rebuild
CM07/CM!2 in that time frame and still meet all SLA.