I have set up a 3-node, multi-subnet Always On Availability group in AWS, which seems to be working well. However, each node only has a single network adaptor with 3 IP addresses assigned to it (1 primary, 1 WSFC IP resource, 1 listener). Is this sufficient for high-availability, or do I need additional networks? I believe with a traditional cluster you would need multiple networks & adaptors to ensure there is no single point of failure. I'm not sure if this is the case in AWS, as I assume they have redundancy built-in at their end?
SQL Server – Always On Availability Groups Network Configuration in AWS
availability-groupsawsNetworksql server
Related Question
- Sql-server – Availability groups NIC best practices
- Sql-server – adding node to existing availability group
- Sql-server – Always On Failover Clustering vs Always On Availability Groups
- SQL Server – Automatic Failover Not Triggered in Always On Availability Groups
- SQL Server AGs – Main Principles Depending on WSFC
- Sql-server – Error Failover in Always on high availability groups
Best Answer
In AWS EC2, you don't get the ability to add additional network ADAPTERS to your VM.
You can add additional network INTERFACES, which are just different IPs/subnets/routes/etc, but they're using the same underlying network adapters (cards).
You can choose to use instance types with faster networking ports, though. The EC2 instance configuration page lists which ones have 10 Gigabit ports, although I like ec2instances.info for easier sorting and filtering. When you're doing Always On Availability Groups, it's especially important to use this faster networking because high-network operations (like backups to another machine) can result in SQL Server timeouts (disclaimer: that's my blog post) if you've only got GbE networking. (That causes AG failovers.)