The main use case for Azure Site Recovery is for disaster recovery. Site Recovery has been in use for almost five to six years. We use these services for DR configuration and disaster recovery.
Replication happens when you have a primary site. The primary site will be replicated to the other side with minimal downtime. We can transfer the data from one region to another region when any disaster happens. That is the reason Microsoft provides us with a 99.95% SLA.
If you have a physical server in your city or country and any disaster happens, you have to switch from your primary to secondary site. If you have hardware, you also have to pay for cooling and resource charges. In Azure, nothing is there; you simply create the DR site, enable replication from primary to secondary. Whenever you require, you can do a quarterly, half-yearly, or monthly failover test as per your business or project requirement.
There is an option to install the agent, establish the communication from your on-premises to cloud, and you can use Azure Site Recovery as well. A major benefit is that you do not want to pay any more for huge costs to build a DR site. The secondary option is the Recovery Services Vault.
Recovery Services Vault will act the same, but there is no way to perform the DR drills. Recovery Services Vault only takes the backup of your virtual machines.
If you set up a physical device, you obviously have to purchase physical devices and resources, but here you only hire one administrator who can configure it without making any noise or taking support from multiple resources.
I am not sure about these improvements because I have been using these services for a long period. It is quite good, and I have not found any negative thoughts on that service.