> On Jun 20, 2018, at 4:02 PM, Zhitao Li <zhitaoli...@gmail.com> wrote: > > Hi, > > We have been working on optimizing container launch latency in our Mesos > based stack,
How are you measuring the launch latency? > and one of the optimization we are considering is to pre-fetch docker image > and any necessary resources for the task/executor. > > This is especially useful in "updating" of containers of long running > services. > > Before delving into detailed proposal, I wonder if anyone has done similar > things or has similar requirements. > > Thanks! > > -- > Cheers, > > Zhitao Li