Re: Oracle RAC with Active Data Guard : Maximum Availability Architecture
From: Chris King <ckaj111_at_yahoo.ca>
Date: Mon, 23 Dec 2013 11:09:33 -0800 (PST)
Message-ID: <1387825773.93853.YahooMailNeo_at_web141405.mail.bf1.yahoo.com>
Thanks all for your feedback. To answer your question, Nassyam Basha, I am looking for 99.5% availability for the RAC instance only. It's okay if the active standby is down outside that timeframe. On Monday, December 23, 2013 11:30:04 AM, Jeremy Schneider <jeremy.schneider_at_ardentperf.com> wrote: 7 (days) x 24 (hours) x 60 (min) x 0.005 (% downtime allowed) = 50 minutes 24 seconds -- per week. I second Andrew. Data Guard is included in enterprise edition and will accomplish what you need. If you're running standard edition or some other edition of oracle then check out dbvisit, that should get you to 90.5 easily. -- http://about.me/jeremy_schneider On Mon, Dec 23, 2013 at 10:21 AM, Andrew Kerber <andrew.kerber_at_gmail.com> wrote: Hmm, 99.5 percent availability isnt that hard these days. A standby should be more than sufficient. At 99.5 percent, that gives you roughly 45 minutes of downtime per week if my math is correct. A switchover or failover to a standby can be done in much less time than that. > > > > >On Mon, Dec 23, 2013 at 9:32 AM, Jeremy Schneider <jeremy.schneider_at_ardentperf.com> wrote: > >Just one thought... keep in mind that "MAA" is also a marketing term. In practice it ends up being sortof synonymous with "maximum complexity architecture" and "maximum licensing architecture"... so I think you're wise to raise the question of complexity. I've been in plenty of situations where RAC and DG were good architectural decision (and the systems I'm work on now are RAC+DG aka MAA) - but there are many factors to availability and I would recommend asking lots of questions and keeping a healthy skepticism! >> >> >>We've had downtime situations this year which would not have happened if we hadn't moved to RAC - including both software bugs in cluster-specific modules and operator errors directly due to new complexity. On the other hand, we've also had situations where downtime was reduced because critical services stayed up during node failures. And that enabled us to schedule maintenance of less-critical broken services for a maintenance window, which was impossible before moving to RAC. >> >> >>-Jeremy >> >> >>-- >>http://about.me/jeremy_schneider >> >> >> >> >>On Mon, Dec 23, 2013 at 8:37 AM, Chris King <ckaj111_at_yahoo.ca> wrote: >> >>We're architecting a new system, and will need 99.5% availability. Looking over Oracle's MAA options, I see RAC with Data Guard. .. but it's not clear to me how this would actually work to avoid ALL downtime. Won't I still need downtime to apply patches, even if I use rolling patches? And isn't it more complex applying patches when there's a standby involved? >>> >>> >>> >>>Is 99.5% really achievable with this combination? >>> >>> >>>Thanks in advance.. >>> >> > > >-- >Andrew W. Kerber > >'If at first you dont succeed, dont take up skydiving.'
Date: Mon, 23 Dec 2013 11:09:33 -0800 (PST)
Message-ID: <1387825773.93853.YahooMailNeo_at_web141405.mail.bf1.yahoo.com>
Thanks all for your feedback. To answer your question, Nassyam Basha, I am looking for 99.5% availability for the RAC instance only. It's okay if the active standby is down outside that timeframe. On Monday, December 23, 2013 11:30:04 AM, Jeremy Schneider <jeremy.schneider_at_ardentperf.com> wrote: 7 (days) x 24 (hours) x 60 (min) x 0.005 (% downtime allowed) = 50 minutes 24 seconds -- per week. I second Andrew. Data Guard is included in enterprise edition and will accomplish what you need. If you're running standard edition or some other edition of oracle then check out dbvisit, that should get you to 90.5 easily. -- http://about.me/jeremy_schneider On Mon, Dec 23, 2013 at 10:21 AM, Andrew Kerber <andrew.kerber_at_gmail.com> wrote: Hmm, 99.5 percent availability isnt that hard these days. A standby should be more than sufficient. At 99.5 percent, that gives you roughly 45 minutes of downtime per week if my math is correct. A switchover or failover to a standby can be done in much less time than that. > > > > >On Mon, Dec 23, 2013 at 9:32 AM, Jeremy Schneider <jeremy.schneider_at_ardentperf.com> wrote: > >Just one thought... keep in mind that "MAA" is also a marketing term. In practice it ends up being sortof synonymous with "maximum complexity architecture" and "maximum licensing architecture"... so I think you're wise to raise the question of complexity. I've been in plenty of situations where RAC and DG were good architectural decision (and the systems I'm work on now are RAC+DG aka MAA) - but there are many factors to availability and I would recommend asking lots of questions and keeping a healthy skepticism! >> >> >>We've had downtime situations this year which would not have happened if we hadn't moved to RAC - including both software bugs in cluster-specific modules and operator errors directly due to new complexity. On the other hand, we've also had situations where downtime was reduced because critical services stayed up during node failures. And that enabled us to schedule maintenance of less-critical broken services for a maintenance window, which was impossible before moving to RAC. >> >> >>-Jeremy >> >> >>-- >>http://about.me/jeremy_schneider >> >> >> >> >>On Mon, Dec 23, 2013 at 8:37 AM, Chris King <ckaj111_at_yahoo.ca> wrote: >> >>We're architecting a new system, and will need 99.5% availability. Looking over Oracle's MAA options, I see RAC with Data Guard. .. but it's not clear to me how this would actually work to avoid ALL downtime. Won't I still need downtime to apply patches, even if I use rolling patches? And isn't it more complex applying patches when there's a standby involved? >>> >>> >>> >>>Is 99.5% really achievable with this combination? >>> >>> >>>Thanks in advance.. >>> >> > > >-- >Andrew W. Kerber > >'If at first you dont succeed, dont take up skydiving.'
-- http://www.freelists.org/webpage/oracle-lReceived on Mon Dec 23 2013 - 20:09:33 CET