My 128GB VMs take 2 hours to back up and 1.5 hours to restore. Is there a way to decrease that time?
ASR has a few potential bottlenecks. Whether or not you can speed it up depends on where this is.
If your on-prem process server is low on resources, that can cause slow replication. Identify what resources are being taxed and increase them (disk IO, RAM, network)
Your WAN may be slow. Quick napkin math puts 128GB at 2 hours at just under 20Mbps. Do you have greater bandwidth available?
You may be doing a full sync and failback. During an initial seed, all data is copied. Subsequently, only changed blocks are replicated out. This is normal, and the initial replication takes far longer than other replication activities because of it.
You can also use something like a virtual Riverbed SteelHead WAN optimizer to improve the speed of the initial seed data. Microsoft wrote a whitepaper on this subject that might be worth reviewing: https://www.microsoft.com/en-us/download/details.aspx?id=44571