ia64/xen-unstable

changeset 10208:6d476981e3a5

Update docs for new credit scheduler and cleanup out of date scheduler stuff.
Signed-off-by: Emmanuel Ackaouy <ack@xensource.com>
author ack@kneesa.uk.xensource.com
date Sun May 28 15:49:17 2006 +0100 (2006-05-28)
parents 444abe25611f
children d5f98d23427a
files docs/src/interface.tex docs/src/user.tex
line diff
     1.1 --- a/docs/src/interface.tex	Sun May 28 09:46:30 2006 +0100
     1.2 +++ b/docs/src/interface.tex	Sun May 28 15:49:17 2006 +0100
     1.3 @@ -205,30 +205,23 @@ event sent at a specified system time by
     1.4  implement timeout values when they block.
     1.5  
     1.6  
     1.7 -
     1.8 -%% % akw: demoting this to a section -- not sure if there is any point
     1.9 -%% % though, maybe just remove it.
    1.10 -
    1.11 -% KAF: Remove these random sections!
    1.12 -\begin{comment}
    1.13  \section{Xen CPU Scheduling}
    1.14  
    1.15  Xen offers a uniform API for CPU schedulers.  It is possible to choose
    1.16  from a number of schedulers at boot and it should be easy to add more.
    1.17 -The BVT, Atropos and Round Robin schedulers are part of the normal Xen
    1.18 -distribution.  BVT provides proportional fair shares of the CPU to the
    1.19 -running domains.  Atropos can be used to reserve absolute shares of
    1.20 -the CPU for each domain.  Round-robin is provided as an example of
    1.21 -Xen's internal scheduler API.
    1.22 +The SEDF, BVT, and Credit schedulers are part of the normal Xen
    1.23 +distribution.  BVT and SEDF will be going away and their use should be
    1.24 +avoided once the credit scheduler has stabilized and become the default.
    1.25 +The Credit scheduler provides proportional fair shares of the
    1.26 +host's CPUs to the running domains. It does this while transparently
    1.27 +load balancing runnable VCPUs across the whole system.
    1.28  
    1.29  \paragraph*{Note: SMP host support}
    1.30 -Xen has always supported SMP host systems.  Domains are statically
    1.31 -assigned to CPUs, either at creation time or when manually pinning to
    1.32 -a particular CPU.  The current schedulers then run locally on each CPU
    1.33 -to decide which of the assigned domains should be run there. The
    1.34 -user-level control software can be used to perform coarse-grain
    1.35 -load-balancing between CPUs.
    1.36 -\end{comment}
    1.37 +Xen has always supported SMP host systems. When using the credit scheduler,
    1.38 +a domain's VCPUs will be dynamically moved across physical CPUs to maximise
    1.39 +domain and system throughput. VCPUs can also be manually restricted to be
    1.40 +mapped only on a subset of the host's physical CPUs, using the pinning
    1.41 +mechanism.
    1.42  
    1.43  
    1.44  %% More information on the characteristics and use of these schedulers
     2.1 --- a/docs/src/user.tex	Sun May 28 09:46:30 2006 +0100
     2.2 +++ b/docs/src/user.tex	Sun May 28 15:49:17 2006 +0100
     2.3 @@ -1094,6 +1094,36 @@ running domains in \xend's SXP configura
     2.4  You can get access to the console of a particular domain using 
     2.5  the \verb_# xm console_ command  (e.g.\ \verb_# xm console myVM_). 
     2.6  
     2.7 +\subsection{Domain Scheduling Management Commands}
     2.8 +
     2.9 +The credit CPU scheduler automatically load balances guest VCPUs
    2.10 +across all available physical CPUs on an SMP host. The user need
    2.11 +not manually pin VCPUs to load balance the system. However, she
    2.12 +can restrict which CPUs a particular VCPU may run on using
    2.13 +the \path{xm vcpu-pin} command.
    2.14 +
    2.15 +Each guest domain is assigned a \path{weight} and a \path{cap}.
    2.16 +
    2.17 +A domain with a weight of 512 will get twice as much CPU as a
    2.18 +domain with a weight of 256 on a contended host. Legal weights
    2.19 +range from 1 to 65535 and the default is 256.
    2.20 +
    2.21 +The cap optionally fixes the maximum amount of CPU a guest will
    2.22 +be able to consume, even if the host system has idle CPU cycles.
    2.23 +The cap is expressed in percentage of one physical CPU: 100 is
    2.24 +1 physical CPU, 50 is half a CPU, 400 is 4 CPUs, etc... The
    2.25 +default, 0, means there is no upper cap.
    2.26 +
    2.27 +When you are running with the credit scheduler, you can check and
    2.28 +modify your domains' weights and caps using the \path{xm sched-credit}
    2.29 +command:
    2.30 +
    2.31 +\begin{tabular}{ll}
    2.32 +\verb!xm sched-credit -d <domain>! & lists weight and cap \\
    2.33 +\verb!xm sched-credit -d <domain> -w <weight>! & sets the weight \\
    2.34 +\verb!xm sched-credit -d <domain> -c <cap>! & sets the cap
    2.35 +\end{tabular}
    2.36 +
    2.37  
    2.38  
    2.39  %% Chapter Domain Configuration
    2.40 @@ -1985,7 +2015,7 @@ editing \path{grub.conf}.
    2.41  \item [ tbuf\_size=xxx ] Set the size of the per-cpu trace buffers, in
    2.42    pages (default 0).  
    2.43  \item [ sched=xxx ] Select the CPU scheduler Xen should use.  The
    2.44 -  current possibilities are `sedf' (default) and `bvt'.
    2.45 +  current possibilities are `sedf' (default), `credit', and `bvt'.
    2.46  \item [ apic\_verbosity=debug,verbose ] Print more detailed
    2.47    information about local APIC and IOAPIC configuration.
    2.48  \item [ lapic ] Force use of local APIC even when left disabled by