Recovery of JIRA index snapshot is very slow

I have a large JIRA 6.2.7 instance that creates a 3.5GB index snapshot once a day. When I try to use the index snapshot on another failover server the import is very slow. From the attached screenshot it took 34 minutes to import 20% of the index. So a full import looks like it would take nearly 3 hours. I can rebuild the index from scratch in about 3.5 hours.

My understanding is that one major purpose of index snapshots is to make failover times shorter. That is not happening for my large JIRA instance (1.2M issues).

As a possible workaround, is it possible to unzip the index snapshot directly into the caches directory and then restart JIRA? It might be quicker using a command line tool rather than unzipping inside the JVM