I'm currently looking to implement a similar setup. My understanding is that an AlwaysOn setup in Azure would be an extension of the current setup, if you were wanting a hybrid type situation. I initially was going to have the Azure SQL Server VM be the primary, however, in my scenario I cannot do this due to customer requirements. They require the ability to still keep functioning if a site goes offline (no Internet). I suppose, I should redact that statement and state that that setup may not be the best in given the requirements.
However, there are some pros to this setup, especially if done correctly. Depending on the SLA you're looking for, geo-replicating this setup to a data center on a another coast is recommended as it will allow for an entire coast DC to go down and you may still remain operational.
PROS (to name a few):
- Easy to geo-replicate
- Decrease in maintenance costs
- Decrease in expenses related to additional servers, even if they are VMs. You're looking at possibly adding additional vCPUs, RAM and possibly storage etc.
- Elastic compute options which will allow for an easy increase in performance, storage etc. and may be changed at any time.
- Can manage those servers as you would an on-prem and from SSMS.
CONS (to name a few):
- There is a learning curve
- Challenges related to stakeholders, management etc. about the "Cloud Stigma".
- Requires additional configuration and setup initially which may be daunting at first. But once setup it doesn't really need to change unless the business processes change etc.
- An S2S VPN tunnel is required, which would be part of the configuration step above. I added as its on con due to the complexity of the setup. Complexity mainly due to Azure terminology.