Information Technology Reference
In-Depth Information
The
zonecfg
command is the most useful tool to set resource controls on
Containers, but its settings do not take effect until the next time the Container
boots. Other useful commands include
prctl
(1) and
rcapadm
(1M), both of which
have an immediate effect on a Container. Resource control settings can be viewed
with the commands
prctl
,
rcapstat
(1), and
poolstat
(1M).
Oracle Solaris 10 includes a wide variety of tools that report on resource
consumption, including
prstat
,
poolstat
(1M),
mpstat
(1M),
rcapstat
, and
kstat
(1M). Many of these features can be applied to Containers using command-
line options or other methods. In addition, a few companies produce excellent
software products designed for monitoring resource consumption by the global
zone and by individual Containers—for example, Oracle Enterprise Manager Ops
Center, BMC Patrol, CA Unicenter, HP Operations Manager, TeamQuest, and oth-
ers. This section discusses only those tools that are included with Oracle Solaris
10. Chapter 9, “Virtualization Management,” discusses Ops Center.
One of the greatest strengths of Containers is common to most OSV imple-
mentations but absent from other forms of virtualization—namely, the ability of
tools running in the management area to provide a holistic view of system activity
(all VEs) as well as a detailed view of activity in each VE. The monitoring tools
described in this section are common Solaris tools that provide either a holistic
view, a detailed view, or options to do both.
6.2.2.1 CPU Controls
Oracle Solaris offers three methods of controlling a Container's use of CPU time.
First, the Fair Share Scheduler is the most flexible, allowing multiple Containers
to share the system's CPUs, or a subset of them. Second, Dynamic Resource Pools
create subsets of CPUs, and allow you to exclusively assign a Container to a subset
called a
pool
. Third, CPU caps enforce a maximum amount of computational abil-
ity on a Container. Although you can use a Container without any CPU controls,
we recommend using these controls to improve performance predictability.
Fair Share Scheduler
The best all-around mechanism is the Fair Share Scheduler
(FSS), which was discussed earlier in Chapter 1. FSS can control the allocation
of available CPU resources among workloads based on their importance. This
importance is expressed by the number of
cpu-shares
that you assign to each
workload. The FSS compares the number of shares assigned to a particular workload
to the aggregate number of shares assigned to all workloads. If one workload has
100 shares and the total number of shares assigned to all workloads is 500, the
scheduler will ensure that the workload receives one-fifth of the available CPU
resources at a minimum.
The last point is very important. A key advantage of this CPU control mecha-
nism over the other options is that it does not waste idle CPU cycles. With FSS,
Search WWH ::
Custom Search