Console doesn't sleep. Most of the tuning here is probably how efficient various steps are at the node level and so probably require code changes.
With Wordpress, the source has to be downloaded and extracted (net and disk io bound), and then the cart is started. Some of the time could be cut by doing steps in parallel, but that again might dip into node changes.
Have you altered your cgroups settings to give more CPU time to the cartridge?
These times seem reasonable until we have time to start making more dramatic optimizations to some core node behavior.
On Jun 3, 2013, at 11:13 AM, Matt Hicks <mhicks redhat com> wrote: