March 27, 2015 at 6:05 am
Hi everyone
We have 2 servers that have almost identical hardware. Currently only 1 is being utilized as a production server.
We would like to setup a High Availability/Always On environment. I have had a look at Failover Clustering and Availability Groups. I am leaning more towards AG as we can use the optimised local RAID drives on each server and there wont be a loss in connection if server A goes offline.
As I understand it:
Failover: Needs to have shared storage, will be downtime while the instance is being transfered
AG: Can use local storage and there will be no downtime if the primary server goes down
Is my thinking correct and is AG the way to go in this scenario?
Thank you.
March 27, 2015 at 6:26 am
Jako de Wet (3/27/2015)
I have had a look at Failover Clustering and Availability Groups. I am leaning more towards AG as we can use the optimised local RAID drives on each server and there wont be a loss in connection if server A goes offline.
Depends if the replica sessions are synchronous or asynchronous
Jako de Wet (3/27/2015)
Failover: Needs to have shared storage, will be downtime while the instance is being transferred
Not entirely correct
A failover cluster instance of sql server requires shared storage and downtime will be encountered during a failover.
AlwaysOn groups use windows server failover clusters, the same technology that sits under a failover cluster instance of SQL Server. During a failover between AlwaysOn group replicas, downtime may still be encountered on synchronous or asynchronous sessions.
Jako de Wet (3/27/2015)
AG: Can use local storage and there will be no downtime if the primary server goes down.
Not quite, see above.
Have a look through my stairway series on this site starting at this link
http://www.sqlservercentral.com/articles/FCI/107536/[/url]
This will give you a better understanding of
-----------------------------------------------------------------------------------------------------------
"Ya can't make an omelette without breaking just a few eggs" 😉
March 27, 2015 at 6:30 am
Thank you for the information Perry.
Will go have a look at your article.
March 27, 2015 at 6:47 am
Perry, after reading the first article, I feel that AO will suit us better for the following reasons:
1. We can use the optimised RAID drives configured on each server, our SAN config is slower atm.
2. We can still use the Instance on server B to create and work with databases not part of the AG
Going to go through the whole stairway though, but I think this is the right option?
I am aware that there might still be a delay in access when the primary goes offline.
March 27, 2015 at 6:54 am
read the whole stairway before making any firm decision 😉
-----------------------------------------------------------------------------------------------------------
"Ya can't make an omelette without breaking just a few eggs" 😉
March 27, 2015 at 8:49 am
Jako, I have NEVER heard of someone reading a forum post or blog series, slapping together a windows failover cluster and dropping SQL Server AG(s) on them and have it work out properly for them. Sometimes it is downright disastrous. Assuming you are doing this for your company's HA/DR plan PLEASE note that if you mess it up the company's continued existence is on the line.
Best,
Kevin G. Boles
SQL Server Consultant
SQL MVP 2007-2012
TheSQLGuru on googles mail service
March 27, 2015 at 8:53 am
Hi Kevin
Dont worry, we have another company that do our server setups, I want to know what direction to send them in.
March 27, 2015 at 9:47 am
Jako de Wet (3/27/2015)
Hi KevinDont worry, we have another company that do our server setups, I want to know what direction to send them in.
The only way I know of to have no disconnect on a server failure is one of the 2-phase-compuational stacks. There used to be 2 companies that provided that (one was Marathon, can't remember the other), but I have no idea if they even exists these days.
Also note there are LOT of things to consider when thinking about HA/DR - WAY more than just your SQL Server.
Best,
Kevin G. Boles
SQL Server Consultant
SQL MVP 2007-2012
TheSQLGuru on googles mail service
March 27, 2015 at 10:01 am
Hi Kevin
Thanks for the input. We are still busy with research before we implement any HA solution. We are ok with a little downtime, but want to decrease the time spent switching over to a new box, should a catastrophic failure occur.
Our server people are pretty clued up and will make sure that they take all into consideration and also run this in a development environment first.
Viewing 9 posts - 1 through 8 (of 8 total)
You must be logged in to reply to this topic. Login to reply