In order to truly understand what are the driving factors in the initial sync time, we need to track more metrics on specific aspects of the initial sync process.

Other dimensions that could be important, such as -

• how many indexes were there?
• Are the indexes large / $text / compound with many keys?
• What was the disk performance like? Is their workload insert heavy?

We need to think through all of the areas we believe could be driving initial sync times and see if we can quantify and track them through different metrics.