Senior DBA Class How to Design Quorum for Failover Clusters
HTML-код
- Опубликовано: 13 мар 2021
- Before you touch anything with Always On in the name, you need to understand how cluster quorum works and how it can bring your SQL Server down. Learn how dynamic quorum and dynamic witness should be configured and see 3 real-world scenarios.
Наука
Thanks for publishing these videos Brent , a good watch
You're welcome!
Hi Brent. Great video! Thanks for all you do for us. At the 13:42 mark you state that a cloud witness is not something we want to use in SQL Server. I’m interested in understanding your reasoning and experiences behind this statement. Unfortunately as DBAs sometimes we have to work on systems that we don’t entirely control or collaborate with other teams on so it may actually be safer to set the witness up in the cloud provided that there is a redundant internet connection than to have it on a local file share on a server that gets arbitrarily rebooted during patching , etc...
Why would you be confident that the network wouldn't also undergo maintenance?
My biggest concern with cloud witness is more on security. It's an amazing feature, no doubt about it. But are you really going to allow traffic to and from port 443 from the public internet to your failover clusters? The reason most customers leverage cloud witness is mainly for multi-datacenter deployments of failover clusters. This means you want to enable automatic failover between data centers. Is that really what you want?
Hi Brent, at 21:50 asynchronous replica /for offloading reads/ shouldn't be in DR data center? Thanks.
It can be, but it doesn't have to be.
Thanks! these videos have been really helpful. we are thinking about adding a two node Failover Cluster in Secondary with Availability groups on a two node Failover Cluster in Primary. For Quorum configuration, our plan is to set up the file share witness in primary data center and when failed over to DR, then modify quorum witness with adjusting node votes? is there another way to accomplish this without having to modify the quorum when we failover to secondary data center?
Architecture of something that complex isn't really a good fit for a RUclips comment.
A common mistake that I see people make with these types of architecture is they only look at the technology portion without looking at the process and people portions. How sure are you that you will be available when a disaster happens that you need to fail over to DR? Can you guarantee that the people who will initiate DR know exactly what to do? And if they don't, would you be willing to be on-call 24/7?
Is this the same for Windows Server 2016?
This class was recorded before Windows Server 2016, and it's just left free online for posterity. For information about newer versions, your best bet is to attend newer training classes.