initial

[rrdtool-all.git] / program / doc / rrdtool.pod

=head1 NAME

rrdtool - Round Robin Database Tool

=head1 SYNOPSIS

B<rrdtool> B<-> [workdir]| I<function>

=head1 DESCRIPTION

=head2 OVERVIEW

It is pretty easy to gather status information from all sorts of
things, ranging from the temperature in your office to the number of
octets which have passed through the FDDI interface of your
router. But it is not so trivial to store this data in an efficient and
systematic manner. This is where B<RRDtool> comes in handy. It lets you
I<log and analyze> the data you gather from all kinds of data-sources
(B<DS>). The data analysis part of RRDtool is based on the ability to
quickly generate graphical representations of the data values
collected over a definable time period.

In this man page you will find general information on the design and
functionality of the Round Robin Database Tool (RRDtool). For a more
detailed description of how to use the individual functions of
B<RRDtool> check the corresponding man page.

For an introduction to the usage of RRDtool make sure you consult the
L<rrdtutorial>.

=head2 FUNCTIONS

While the man pages talk of command line switches you have to set in
order to make B<RRDtool> work it is important to note that
B<RRDtool> can be remotely controlled through a set of pipes. This
saves a considerable amount of startup time when you plan to make
B<RRDtool> do a lot of things quickly. Check the section on L<Remote_Control>
further down. There is also a number of language bindings
for RRDtool which allow you to use it directly from Perl, python, Tcl,
PHP, etc.

=over 8

=item B<create>

Set up a new Round Robin Database (RRD). Check L<rrdcreate>.

=item B<update>

Store new data values into an RRD. Check L<rrdupdate>.

=item B<updatev>

Operationally equivalent to B<update> except for output. Check L<rrdupdate>.

=item B<graph>

Create a graph from data stored in one or several RRDs. Apart from
generating graphs, data can also be extracted to stdout. Check L<rrdgraph>.

=item B<dump>

Dump the contents of an RRD in plain ASCII. In connection with restore
you can use this to move an RRD from one computer architecture to
another.  Check L<rrddump>.

=item B<restore>

Restore an RRD in XML format to a binary RRD. Check L<rrdrestore>

=item B<fetch>

Get data for a certain time period from a RRD. The graph function
uses fetch to retrieve its data from an RRD. Check L<rrdfetch>.

=item B<tune>

Alter setup of an RRD. Check L<rrdtune>.

=item B<last>

Find the last update time of an RRD. Check L<rrdlast>.

=item B<info>

Get information about an RRD. Check L<rrdinfo>.

=item B<rrdresize>

Change the size of individual RRAs. This is dangerous! Check L<rrdresize>.

=item B<xport>

Export data retrieved from one or several RRDs. Check L<rrdxport>.

=item B<flushcached>

Flush the values for a specific RRD file from memory. Check L<rrdflushcached>.

=item B<rrdcgi>

This is a standalone tool for producing RRD graphs on the fly. Check
L<rrdcgi>.

=back

=head2 HOW DOES RRDTOOL WORK?

=over 8

=item Data Acquisition

When monitoring the state of a system, it is convenient to have the
data available at a constant time interval. Unfortunately, you may not
always be able to fetch data at exactly the time you want
to. Therefore B<RRDtool> lets you update the log file at any time you
want. It will automatically interpolate the value of the data-source
(B<DS>) at the latest official time-slot (interval) and write this
interpolated value to the log. The original value you have supplied is
stored as well and is also taken into account when interpolating the
next log entry.

=item Consolidation

You may log data at a 1 minute interval, but you might also be
interested to know the development of the data over the last year. You
could do this by simply storing the data in 1 minute intervals for the
whole year. While this would take considerable disk space it would
also take a lot of time to analyze the data when you wanted to create
a graph covering the whole year. B<RRDtool> offers a solution to this
problem through its data consolidation feature. When setting up an
Round Robin Database (B<RRD>), you can define at which interval this
consolidation should occur, and what consolidation function (B<CF>)
(average, minimum, maximum, total, last) should be used to build the
consolidated values (see rrdcreate). You can define any number of
different consolidation setups within one B<RRD>. They will all be
maintained on the fly when new data is loaded into the B<RRD>.

=item Round Robin Archives

Data values of the same consolidation setup are stored into Round
Robin Archives (B<RRA>). This is a very efficient manner to store data
for a certain amount of time, while using a known and constant amount
of storage space.

It works like this: If you want to store 1'000 values in 5 minute
interval, B<RRDtool> will allocate space for 1'000 data values and a
header area. In the header it will store a pointer telling which slots
(value) in the storage area was last written to. New values are
written to the Round Robin Archive in, you guessed it, a round robin
manner. This automatically limits the history to the last 1'000 values
(in our example). Because you can define several B<RRA>s within a
single B<RRD>, you can setup another one, for storing 750 data values
at a 2 hour interval, for example, and thus keep a log for the last
two months at a lower resolution.

The use of B<RRA>s guarantees that the B<RRD> does not grow over
time and that old data is automatically eliminated. By using the
consolidation feature, you can still keep data for a very long time,
while gradually reducing the resolution of the data along the time
axis.

Using different consolidation functions (B<CF>) allows you to store
exactly the type of information that actually interests you: the maximum
one minute traffic on the LAN, the minimum temperature of your wine cellar,
the total minutes of down time, etc.

=item Unknown Data

As mentioned earlier, the B<RRD> stores data at a constant
interval. Sometimes it may happen that no new data is available when a
value has to be written to the B<RRD>. Data acquisition may not be
possible for one reason or other. With B<RRDtool> you can handle these
situations by storing an I<*UNKNOWN*> value into the database. The
value 'I<*UNKNOWN*>' is supported through all the functions of the
tool. When consolidating a data set, the amount of I<*UNKNOWN*> data
values is accounted for and when a new consolidated value is ready to
be written to its Round Robin Archive (B<RRA>), a validity check is
performed to make sure that the percentage of unknown values in the
data point is above a configurable level. If not, an I<*UNKNOWN*> value
will be written to the B<RRA>.

=item Graphing

B<RRDtool> allows you to generate reports in numerical and
graphical form based on the data stored in one or several
B<RRD>s. The graphing feature is fully configurable. Size, color and
contents of the graph can be defined freely. Check L<rrdgraph>
for more information on this.

=item Aberrant Behavior Detection

by Jake Brutlag

B<RRDtool> provides the building blocks for near real-time aberrant
behavior detection. These components include:

=over

=item *

An algorithm for predicting the value of a time series one time step
into the future.

=item *

A measure of deviation between predicted and observed values.

=item *

A mechanism to decide if and when an observed value or sequence of
observed values is I<too deviant> from the predicted value(s).

=back

Here is a brief explanation of these components:

The Holt-Winters time series forecasting algorithm is an on-line (or
incremental) algorithm that adaptively predicts future observations in
a time series. Its forecast is the sum of three components: a baseline
(or intercept), a linear trend over time (or slope), and a seasonal
coefficient (a periodic effect, such as a daily cycle). There is one
seasonal coefficient for each time point in the period (cycle). After
a value is observed, each of these components is updated via
exponential smoothing. This means that the algorithm "learns" from
past values and uses them to predict the future. The rate of
adaptation is governed by 3 parameters, alpha (intercept), beta
(slope), and gamma (seasonal). The prediction can also be viewed as a
smoothed value for the time series.

The measure of deviation is a seasonal weighted absolute
deviation. The term I<seasonal> means deviation is measured separately
for each time point in the seasonal cycle. As with Holt-Winters
forecasting, deviation is predicted using the measure computed from
past values (but only at that point in the seasonal cycle). After the
value is observed, the algorithm learns from the observed value via
exponential smoothing. Confidence bands for the observed time series
are generated by scaling the sequence of predicted deviation values
(we usually think of the sequence as a continuous line rather than a
set of discrete points).

Aberrant behavior (a potential failure) is reported whenever the
number of times the observed value violates the confidence bands meets
or exceeds a specified threshold within a specified temporal window
(e.g. 5 violations during the past 45 minutes with a value observed
every 5 minutes).

This functionality is embedded in a set of related B<RRAs>. In
particular, a FAILURES B<RRA> logs potential failures. With these data
you could, for example, use a front-end application to B<RRDtool> to
initiate real-time alerts.

For a detailed description on how to set this up, see L<rrdcreate>.

=back

=head2 REMOTE CONTROL

When you start B<RRDtool> with the command line option 'B<->' it waits
for input via standard input (STDIN). With this feature you can
improve performance by attaching B<RRDtool> to another process (MRTG
is one example) through a set of pipes. Over these pipes B<RRDtool>
accepts the same arguments as on the command line and some special
commands like B<quit, cd, mkdir> and B<ls>. For detailed help on the
server commands type:

   rrdtool help cd|mkdir|pwd|ls|quit

When a command is completed, RRDtool will print the string  'C<OK>',
followed by timing information of the form B<u:>I<usertime>
B<s:>I<systemtime>. Both values are the running totals of seconds since
RRDtool was started. If an error occurs, a line of the form 'C<ERROR:>
I<Description of error>' will be printed instead. B<RRDtool> will not abort,
unless something really serious happens. If
a B<workdir> is specified and the UID is 0, RRDtool will do a chroot to that
workdir. If the UID is not 0, RRDtool only changes the current directory to
B<workdir>.

=head2 RRD Server

If you want to create a RRD-Server, you must choose a TCP/IP Service
number and add them to I</etc/services> like this:

 rrdsrv      13900/tcp                       # RRD server

Attention: the TCP port 13900 isn't officially registered for
rrdsrv. You can use any unused port in your services file, but the
server and the client system must use the same port, of course.

With this configuration you can add RRDtool as meta-server to
I</etc/inetd.conf>. For example:

 rrdsrv stream tcp nowait root /opt/rrd/bin/rrdtool rrdtool - /var/rrd

Don't forget to create the database directory /var/rrd and
reinitialize your inetd.

If all was setup correctly, you can access the server with Perl
sockets, tools like netcat, or in a quick interactive test by using
'telnet localhost rrdsrv'.

B<NOTE:> that there is no authentication with this feature! Do not setup
such a port unless you are sure what you are doing.

=head1 RRDCACHED, THE CACHING DAEMON

For very big setups, updating thousands of RRD files often becomes a serious IO
problem. If you run into such problems, you might want to take a look at
L<rrdcached>, a caching daemon for RRDtool which may help you lessen the
stress on your disks.

=head1 SEE ALSO

rrdcreate, rrdupdate, rrdgraph, rrddump, rrdfetch, rrdtune, rrdlast, rrdxport,
rrdflushcached, rrdcached

=head1 BUGS

Bugs? Features!

=head1 AUTHOR

Tobias Oetiker <tobi@oetiker.ch>