Re: Moving to RAC
From: Dave Morgan <oracle_at_1001111.com>
Date: Wed, 04 Mar 2015 07:56:28 -0700
Message-ID: <54F71D1C.1090509_at_1001111.com>
> Date: Wed, 4 Mar 2015 09:56:28 +1100
> From: Stojan Veselinovski <stojan.veselinovski_at_gmail.com>
>
> Agree with Jeremy's sentiments.
> The initial stages you are likely to get less fault tolerance due to
> configuration, bugs, learnings etc, etc, RAC is not a magic HA bullet.
> Make sure you properly design and architect for it.
Date: Wed, 04 Mar 2015 07:56:28 -0700
Message-ID: <54F71D1C.1090509_at_1001111.com>
> Date: Wed, 4 Mar 2015 09:56:28 +1100
> From: Stojan Veselinovski <stojan.veselinovski_at_gmail.com>
>
> Agree with Jeremy's sentiments.
> The initial stages you are likely to get less fault tolerance due to
> configuration, bugs, learnings etc, etc, RAC is not a magic HA bullet.
> Make sure you properly design and architect for it.
I agree, I always figure 6 months to reach the previous SLA and another 6 months "tuning" the cluster to get true HA.
I also believe a 2 node cluster provides very little benefit and a lot of costs. If both nodes are running at 60% load and one fails, the other will bind up very quickly.
4 or more for HA, less if separating application loads, 1 node per load
>
> Make sure you test load and resilience eg. downing nics in bonded
> interfaces or haip, node evictions, losing io channel, etc, etc
+eleventy-million. My favourite is pulling plugs in the data centre. Always watch the face of the storage admin :)
Dave
-- Dave Morgan Senior Consultant, 1001111 Alberta Limited dave.morgan_at_1001111.com 403 399 2442 -- http://www.freelists.org/webpage/oracle-lReceived on Wed Mar 04 2015 - 15:56:28 CET