The current challenge is how to provide the billing
strategy. Each customer will be operating in it's own Solaris
Container, but there is no obvious way to measure the utilisation by
container.
DTRACE is the obvious choice to enable the customer
to create such a billing infrastructure. Simple calculation of the number of CPU-Seconds and read/write IO's by container over a period of time is the desired result.
The uts core exacct usage and recording routines(exacct) and associated
D fbt probes (fbt:genunix::entry) requires getacct(2), putacct(2), and
wracct(2) system calls which is on top of exacctsys to trap into kernel
instead of call back.
Specifically, the fbt provider will do the work
to probe the uts structure such as proc_usage_t,task_usage_t
and flow_usage_t.
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment