check row count in RRAs to be > 0 ... by Sebastian Harl

[rrdtool.git] / doc / rrdcreate.pod

=head1 NAME

rrdcreate - Set up a new Round Robin Database

=head1 SYNOPSIS

B<rrdtool> B<create> I<filename>
S<[B<--start>|B<-b> I<start time>]>
S<[B<--step>|B<-s> I<step>]>
S<[B<DS:>I<ds-name>B<:>I<DST>B<:>I<dst arguments>]>
S<[B<RRA:>I<CF>B<:>I<cf arguments>]>

=head1 DESCRIPTION

The create function of RRDtool lets you set up new Round Robin
Database (B<RRD>) files.  The file is created at its final, full size
and filled with I<*UNKNOWN*> data.

=over 8

=item I<filename>

The name of the B<RRD> you want to create. B<RRD> files should end
with the extension F<.rrd>. However, B<RRDtool> will accept any
filename.

=item B<--start>|B<-b> I<start time> (default: now - 10s)

Specifies the time in seconds since 1970-01-01 UTC when the first
value should be added to the B<RRD>. B<RRDtool> will not accept
any data timed before or at the time specified.

See also AT-STYLE TIME SPECIFICATION section in the
I<rrdfetch> documentation for other ways to specify time.

=item B<--step>|B<-s> I<step> (default: 300 seconds)

Specifies the base interval in seconds with which data will be fed
into the B<RRD>.

=item B<DS:>I<ds-name>B<:>I<DST>B<:>I<dst arguments>

A single B<RRD> can accept input from several data sources (B<DS>),
for example incoming and outgoing traffic on a specific communication
line. With the B<DS> configuration option you must define some basic
properties of each data source you want to store in the B<RRD>.

I<ds-name> is the name you will use to reference this particular data
source from an B<RRD>. A I<ds-name> must be 1 to 19 characters long in
the characters [a-zA-Z0-9_].

I<DST> defines the Data Source Type. The remaining arguments of a
data source entry depend on the data source type. For GAUGE, COUNTER,
DERIVE, and ABSOLUTE the format for a data source entry is:

B<DS:>I<ds-name>B<:>I<GAUGE | COUNTER | DERIVE | ABSOLUTE>B<:>I<heartbeat>B<:>I<min>B<:>I<max>

For COMPUTE data sources, the format is:

B<DS:>I<ds-name>B<:>I<COMPUTE>B<:>I<rpn-expression>

In order to decide which data source type to use, review the
definitions that follow. Also consult the section on "HOW TO MEASURE"
for further insight.

=over 4

=item B<GAUGE>

is for things like temperatures or number of people in a room or the
value of a RedHat share.

=item B<COUNTER>

is for continuous incrementing counters like the ifInOctets counter in
a router. The B<COUNTER> data source assumes that the counter never
decreases, except when a counter overflows.  The update function takes
the overflow into account.  The counter is stored as a per-second
rate. When the counter overflows, RRDtool checks if the overflow
happened at the 32bit or 64bit border and acts accordingly by adding
an appropriate value to the result.

=item B<DERIVE>

will store the derivative of the line going from the last to the
current value of the data source. This can be useful for gauges, for
example, to measure the rate of people entering or leaving a
room. Internally, derive works exactly like COUNTER but without
overflow checks. So if your counter does not reset at 32 or 64 bit you
might want to use DERIVE and combine it with a MIN value of 0.

=over

=item NOTE on COUNTER vs DERIVE

by Don Baarda E<lt>don.baarda@baesystems.comE<gt>

If you cannot tolerate ever mistaking the occasional counter reset for a
legitimate counter wrap, and would prefer "Unknowns" for all legitimate
counter wraps and resets, always use DERIVE with min=0. Otherwise, using
COUNTER with a suitable max will return correct values for all legitimate
counter wraps, mark some counter resets as "Unknown", but can mistake some
counter resets for a legitimate counter wrap.

For a 5 minute step and 32-bit counter, the probability of mistaking a
counter reset for a legitimate wrap is arguably about 0.8% per 1Mbps of
maximum bandwidth. Note that this equates to 80% for 100Mbps interfaces, so
for high bandwidth interfaces and a 32bit counter, DERIVE with min=0 is
probably preferable. If you are using a 64bit counter, just about any max
setting will eliminate the possibility of mistaking a reset for a counter
wrap.

=back

=item B<ABSOLUTE>

is for counters which get reset upon reading. This is used for fast counters
which tend to overflow. So instead of reading them normally you reset them
after every read to make sure you have a maximum time available before the
next overflow. Another usage is for things you count like number of messages
since the last update.

=item B<COMPUTE>

is for storing the result of a formula applied to other data sources
in the B<RRD>. This data source is not supplied a value on update, but
rather its Primary Data Points (PDPs) are computed from the PDPs of
the data sources according to the rpn-expression that defines the
formula. Consolidation functions are then applied normally to the PDPs
of the COMPUTE data source (that is the rpn-expression is only applied
to generate PDPs). In database software, such data sets are referred
to as "virtual" or "computed" columns.

=back

I<heartbeat> defines the maximum number of seconds that may pass
between two updates of this data source before the value of the
data source is assumed to be I<*UNKNOWN*>.

I<min> and I<max> define the expected range values for data supplied by a
data source. If I<min> and/or I<max> any value outside the defined range
will be regarded as I<*UNKNOWN*>. If you do not know or care about min and
max, set them to U for unknown. Note that min and max always refer to the
processed values of the DS. For a traffic-B<COUNTER> type DS this would be
the maximum and minimum data-rate expected from the device.

I<If information on minimal/maximal expected values is available,
always set the min and/or max properties. This will help RRDtool in
doing a simple sanity check on the data supplied when running update.>

I<rpn-expression> defines the formula used to compute the PDPs of a
COMPUTE data source from other data sources in the same <RRD>. It is
similar to defining a B<CDEF> argument for the graph command. Please
refer to that manual page for a list and description of RPN operations
supported. For COMPUTE data sources, the following RPN operations are
not supported: COUNT, PREV, TIME, and LTIME. In addition, in defining
the RPN expression, the COMPUTE data source may only refer to the
names of data source listed previously in the create command. This is
similar to the restriction that B<CDEF>s must refer only to B<DEF>s
and B<CDEF>s previously defined in the same graph command.

=item B<RRA:>I<CF>B<:>I<cf arguments>


The purpose of an B<RRD> is to store data in the round robin archives
(B<RRA>). An archive consists of a number of data values or statistics for
each of the defined data-sources (B<DS>) and is defined with an B<RRA> line.

When data is entered into an B<RRD>, it is first fit into time slots
of the length defined with the B<-s> option, thus becoming a I<primary
data point>.

The data is also processed with the consolidation function (I<CF>) of
the archive. There are several consolidation functions that
consolidate primary data points via an aggregate function: B<AVERAGE>,
B<MIN>, B<MAX>, B<LAST>. 

=over

=item AVERAGE

the average of the data points is stored.

=item MIN

the smallest of the data points is stored.

=item MAX

the largest of the data points is stored.

=item LAST

the last data points is used.

=back

Note that data aggregation inevitably leads to loss of precision and
information. The trick is to pick the aggregate function such that the
I<interesting> properties of your data is kept across the aggregation
process.


The format of B<RRA> line for these
consolidation functions is:

B<RRA:>I<AVERAGE | MIN | MAX | LAST>B<:>I<xff>B<:>I<steps>B<:>I<rows>

I<xff> The xfiles factor defines what part of a consolidation interval may
be made up from I<*UNKNOWN*> data while the consolidated value is still
regarded as known. It is given as the ratio of allowed I<*UNKNOWN*> PDPs
to the number of PDPs in the interval. Thus, it ranges from 0 to 1 (exclusive).


I<steps> defines how many of these I<primary data points> are used to build
a I<consolidated data point> which then goes into the archive.

I<rows> defines how many generations of data values are kept in an B<RRA>.
Obviously, this has to be greater than zero.

=back

=head1 Aberrant Behavior Detection with Holt-Winters Forecasting

In addition to the aggregate functions, there are a set of specialized
functions that enable B<RRDtool> to provide data smoothing (via the
Holt-Winters forecasting algorithm), confidence bands, and the
flagging aberrant behavior in the data source time series:

=over

=item *

B<RRA:>I<HWPREDICT>B<:>I<rows>B<:>I<alpha>B<:>I<beta>B<:>I<seasonal period>[B<:>I<rra-num>]

=item *

B<RRA:>I<MHWPREDICT>B<:>I<rows>B<:>I<alpha>B<:>I<beta>B<:>I<seasonal period>[B<:>I<rra-num>]

=item *

B<RRA:>I<SEASONAL>B<:>I<seasonal period>B<:>I<gamma>B<:>I<rra-num>[B<:smoothing-window=>I<fraction>]

=item *

B<RRA:>I<DEVSEASONAL>B<:>I<seasonal period>B<:>I<gamma>B<:>I<rra-num>[B<:smoothing-window=>I<fraction>]

=item *

B<RRA:>I<DEVPREDICT>B<:>I<rows>B<:>I<rra-num>

=item *

B<RRA:>I<FAILURES>B<:>I<rows>B<:>I<threshold>B<:>I<window length>B<:>I<rra-num>

=back

These B<RRAs> differ from the true consolidation functions in several ways.
First, each of the B<RRA>s is updated once for every primary data point.
Second, these B<RRAs> are interdependent. To generate real-time confidence
bounds, a matched set of SEASONAL, DEVSEASONAL, DEVPREDICT, and either
HWPREDICT or MHWPREDICT must exist. Generating smoothed values of the primary
data points requires a SEASONAL B<RRA> and either an HWPREDICT or MHWPREDICT 
B<RRA>. Aberrant behavior detection requires FAILURES, DEVSEASONAL, SEASONAL,
and either HWPREDICT or MHWPREDICT.

The predicted, or smoothed, values are stored in the HWPREDICT or MHWPREDICT
B<RRA>. HWPREDICT and MHWPREDICT are actually two variations on the
Holt-Winters method. They are interchangeable. Both attempt to decompose data
into three components: a baseline, a trend, and a seasonal coefficient.
HWPREDICT adds its seasonal coefficient to the baseline to form a prediction, whereas
MHWPREDICT multiplies its seasonal coefficient by the baseline to form a
prediction. The difference is noticeable when the baseline changes
significantly in the course of a season; HWPREDICT will predict the seasonality
to stay constant as the baseline changes, but MHWPREDICT will predict the
seasonality to grow or shrink in proportion to the baseline. The proper choice
of method depends on the thing being modeled. For simplicity, the rest of this
discussion will refer to HWPREDICT, but MHWPREDICT may be substituted in its
place.

The predicted deviations are stored in DEVPREDICT (think a standard deviation
which can be scaled to yield a confidence band). The FAILURES B<RRA> stores 
binary indicators. A 1 marks the indexed observation as failure; that is, the 
number of confidence bounds violations in the preceding window of observations 
met or exceeded a specified threshold. An example of using these B<RRAs> to graph 
confidence bounds and failures appears in L<rrdgraph>.

The SEASONAL and DEVSEASONAL B<RRAs> store the seasonal coefficients for the
Holt-Winters forecasting algorithm and the seasonal deviations, respectively.
There is one entry per observation time point in the seasonal cycle. For
example, if primary data points are generated every five minutes and the
seasonal cycle is 1 day, both SEASONAL and DEVSEASONAL will have 288 rows.

In order to simplify the creation for the novice user, in addition to
supporting explicit creation of the HWPREDICT, SEASONAL, DEVPREDICT,
DEVSEASONAL, and FAILURES B<RRAs>, the B<RRDtool> create command supports
implicit creation of the other four when HWPREDICT is specified alone and
the final argument I<rra-num> is omitted.

I<rows> specifies the length of the B<RRA> prior to wrap around. Remember
that there is a one-to-one correspondence between primary data points and
entries in these RRAs. For the HWPREDICT CF, I<rows> should be larger than
the I<seasonal period>. If the DEVPREDICT B<RRA> is implicitly created, the
default number of rows is the same as the HWPREDICT I<rows> argument. If the
FAILURES B<RRA> is implicitly created, I<rows> will be set to the I<seasonal
period> argument of the HWPREDICT B<RRA>. Of course, the B<RRDtool>
I<resize> command is available if these defaults are not sufficient and the
creator wishes to avoid explicit creations of the other specialized function
B<RRAs>.

I<seasonal period> specifies the number of primary data points in a seasonal
cycle. If SEASONAL and DEVSEASONAL are implicitly created, this argument for
those B<RRAs> is set automatically to the value specified by HWPREDICT. If
they are explicitly created, the creator should verify that all three
I<seasonal period> arguments agree.

I<alpha> is the adaption parameter of the intercept (or baseline)
coefficient in the Holt-Winters forecasting algorithm. See L<rrdtool> for a
description of this algorithm. I<alpha> must lie between 0 and 1. A value
closer to 1 means that more recent observations carry greater weight in
predicting the baseline component of the forecast. A value closer to 0 means
that past history carries greater weight in predicting the baseline
component.

I<beta> is the adaption parameter of the slope (or linear trend) coefficient
in the Holt-Winters forecasting algorithm. I<beta> must lie between 0 and 1
and plays the same role as I<alpha> with respect to the predicted linear
trend.

I<gamma> is the adaption parameter of the seasonal coefficients in the
Holt-Winters forecasting algorithm (HWPREDICT) or the adaption parameter in
the exponential smoothing update of the seasonal deviations. It must lie
between 0 and 1. If the SEASONAL and DEVSEASONAL B<RRAs> are created
implicitly, they will both have the same value for I<gamma>: the value
specified for the HWPREDICT I<alpha> argument. Note that because there is
one seasonal coefficient (or deviation) for each time point during the
seasonal cycle, the adaptation rate is much slower than the baseline. Each
seasonal coefficient is only updated (or adapts) when the observed value
occurs at the offset in the seasonal cycle corresponding to that
coefficient.

If SEASONAL and DEVSEASONAL B<RRAs> are created explicitly, I<gamma> need not
be the same for both. Note that I<gamma> can also be changed via the
B<RRDtool> I<tune> command.

I<smoothing-window> specifies the fraction of a season that should be
averaged around each point. By default, the value of I<smoothing-window> is
0.05, which means each value in SEASONAL and DEVSEASONAL will be occasionally
replaced by averaging it with its (I<seasonal period>*0.05) nearest neighbors.
Setting I<smoothing-window> to zero will disable the running-average smoother
altogether.

I<rra-num> provides the links between related B<RRAs>. If HWPREDICT is
specified alone and the other B<RRAs> are created implicitly, then
there is no need to worry about this argument. If B<RRAs> are created
explicitly, then carefully pay attention to this argument. For each
B<RRA> which includes this argument, there is a dependency between
that B<RRA> and another B<RRA>. The I<rra-num> argument is the 1-based
index in the order of B<RRA> creation (that is, the order they appear
in the I<create> command). The dependent B<RRA> for each B<RRA>
requiring the I<rra-num> argument is listed here:

=over

=item *

HWPREDICT I<rra-num> is the index of the SEASONAL B<RRA>.

=item *

SEASONAL I<rra-num> is the index of the HWPREDICT B<RRA>.

=item *

DEVPREDICT I<rra-num> is the index of the DEVSEASONAL B<RRA>.

=item *

DEVSEASONAL I<rra-num> is the index of the HWPREDICT B<RRA>.

=item *

FAILURES I<rra-num> is the index of the DEVSEASONAL B<RRA>.

=back

I<threshold> is the minimum number of violations (observed values outside
the confidence bounds) within a window that constitutes a failure. If the
FAILURES B<RRA> is implicitly created, the default value is 7.

I<window length> is the number of time points in the window. Specify an
integer greater than or equal to the threshold and less than or equal to 28.
The time interval this window represents depends on the interval between
primary data points. If the FAILURES B<RRA> is implicitly created, the
default value is 9.

=head1 The HEARTBEAT and the STEP

Here is an explanation by Don Baarda on the inner workings of RRDtool.
It may help you to sort out why all this *UNKNOWN* data is popping
up in your databases:

RRDtool gets fed samples/updates at arbitrary times. From these it builds Primary
Data Points (PDPs) on every "step" interval. The PDPs are
then accumulated into the RRAs.

The "heartbeat" defines the maximum acceptable interval between
samples/updates. If the interval between samples is less than "heartbeat",
then an average rate is calculated and applied for that interval. If
the interval between samples is longer than "heartbeat", then that
entire interval is considered "unknown". Note that there are other
things that can make a sample interval "unknown", such as the rate
exceeding limits, or a sample that was explicitly marked as unknown.

The known rates during a PDP's "step" interval are used to calculate
an average rate for that PDP. If the total "unknown" time accounts for
more than B<half> the "step", the entire PDP is marked
as "unknown". This means that a mixture of known and "unknown" sample
times in a single PDP "step" may or may not add up to enough "known"
time to warrent for a known PDP.

The "heartbeat" can be short (unusual) or long (typical) relative to
the "step" interval between PDPs. A short "heartbeat" means you
require multiple samples per PDP, and if you don't get them mark the
PDP unknown. A long heartbeat can span multiple "steps", which means
it is acceptable to have multiple PDPs calculated from a single
sample. An extreme example of this might be a "step" of 5 minutes and a
"heartbeat" of one day, in which case a single sample every day will
result in all the PDPs for that entire day period being set to the
same average rate. I<-- Don Baarda E<lt>don.baarda@baesystems.comE<gt>>

       time|
       axis|
 begin__|00|
        |01|
       u|02|----* sample1, restart "hb"-timer
       u|03|   /
       u|04|  /
       u|05| /
       u|06|/     "hbt" expired
       u|07|
        |08|----* sample2, restart "hb" 
        |09|   / 
        |10|  /
       u|11|----* sample3, restart "hb"
       u|12|   /
       u|13|  /
 step1_u|14| /
       u|15|/     "swt" expired
       u|16|
        |17|----* sample4, restart "hb", create "pdp" for step1 = 
        |18|   /  = unknown due to 10 "u" labled secs > 0.5 * step
        |19|  /
        |20| /
        |21|----* sample5, restart "hb"
        |22|   /
        |23|  /
        |24|----* sample6, restart "hb"
        |25|   /
        |26|  /
        |27|----* sample7, restart "hb"
 step2__|28|   /
        |22|  /
        |23|----* sample8, restart "hb", create "pdp" for step1, create "cdp" 
        |24|   /
        |25|  /

graphics by I<vladimir.lavrov@desy.de>.


=head1 HOW TO MEASURE

Here are a few hints on how to measure:

=over


=item Temperature

Usually you have some type of meter you can read to get the temperature.
The temperature is not really connected with a time. The only connection is
that the temperature reading happened at a certain time. You can use the
B<GAUGE> data source type for this. RRDtool will then record your reading
together with the time.

=item Mail Messages

Assume you have a method to count the number of messages transported by
your mailserver in a certain amount of time, giving you data like '5
messages in the last 65 seconds'. If you look at the count of 5 like an
B<ABSOLUTE> data type you can simply update the RRD with the number 5 and the
end time of your monitoring period. RRDtool will then record the number of
messages per second. If at some later stage you want to know the number of
messages transported in a day, you can get the average messages per second
from RRDtool for the day in question and multiply this number with the
number of seconds in a day. Because all math is run with Doubles, the
precision should be acceptable.

=item It's always a Rate

RRDtool stores rates in amount/second for COUNTER, DERIVE and ABSOLUTE
data.  When you plot the data, you will get on the y axis
amount/second which you might be tempted to convert to an absolute
amount by multiplying by the delta-time between the points. RRDtool
plots continuous data, and as such is not appropriate for plotting
absolute amounts as for example "total bytes" sent and received in a
router. What you probably want is plot rates that you can scale to
bytes/hour, for example, or plot absolute amounts with another tool
that draws bar-plots, where the delta-time is clear on the plot for
each point (such that when you read the graph you see for example GB
on the y axis, days on the x axis and one bar for each day).

=back


=head1 EXAMPLE

 rrdtool create temperature.rrd --step 300 \
  DS:temp:GAUGE:600:-273:5000 \
  RRA:AVERAGE:0.5:1:1200 \
  RRA:MIN:0.5:12:2400 \
  RRA:MAX:0.5:12:2400 \
  RRA:AVERAGE:0.5:12:2400

This sets up an B<RRD> called F<temperature.rrd> which accepts one
temperature value every 300 seconds. If no new data is supplied for
more than 600 seconds, the temperature becomes I<*UNKNOWN*>.  The
minimum acceptable value is -273 and the maximum is 5'000.

A few archive areas are also defined. The first stores the
temperatures supplied for 100 hours (1'200 * 300 seconds = 100
hours). The second RRA stores the minimum temperature recorded over
every hour (12 * 300 seconds = 1 hour), for 100 days (2'400 hours). The
third and the fourth RRA's do the same for the maximum and
average temperature, respectively.

=head1 EXAMPLE 2

 rrdtool create monitor.rrd --step 300        \
   DS:ifOutOctets:COUNTER:1800:0:4294967295   \
   RRA:AVERAGE:0.5:1:2016                     \
   RRA:HWPREDICT:1440:0.1:0.0035:288

This example is a monitor of a router interface. The first B<RRA> tracks the
traffic flow in octets; the second B<RRA> generates the specialized
functions B<RRAs> for aberrant behavior detection. Note that the I<rra-num>
argument of HWPREDICT is missing, so the other B<RRAs> will implicitly be
created with default parameter values. In this example, the forecasting
algorithm baseline adapts quickly; in fact the most recent one hour of
observations (each at 5 minute intervals) accounts for 75% of the baseline
prediction. The linear trend forecast adapts much more slowly. Observations
made during the last day (at 288 observations per day) account for only
65% of the predicted linear trend. Note: these computations rely on an
exponential smoothing formula described in the LISA 2000 paper.

The seasonal cycle is one day (288 data points at 300 second intervals), and
the seasonal adaption parameter will be set to 0.1. The RRD file will store 5
days (1'440 data points) of forecasts and deviation predictions before wrap
around. The file will store 1 day (a seasonal cycle) of 0-1 indicators in
the FAILURES B<RRA>.

The same RRD file and B<RRAs> are created with the following command,
which explicitly creates all specialized function B<RRAs>.

 rrdtool create monitor.rrd --step 300 \
   DS:ifOutOctets:COUNTER:1800:0:4294967295 \
   RRA:AVERAGE:0.5:1:2016 \
   RRA:HWPREDICT:1440:0.1:0.0035:288:3 \
   RRA:SEASONAL:288:0.1:2 \
   RRA:DEVPREDICT:1440:5 \
   RRA:DEVSEASONAL:288:0.1:2 \
   RRA:FAILURES:288:7:9:5

Of course, explicit creation need not replicate implicit create, a
number of arguments could be changed.

=head1 EXAMPLE 3

 rrdtool create proxy.rrd --step 300 \
   DS:Total:DERIVE:1800:0:U  \
   DS:Duration:DERIVE:1800:0:U  \
   DS:AvgReqDur:COMPUTE:Duration,Requests,0,EQ,1,Requests,IF,/ \
   RRA:AVERAGE:0.5:1:2016

This example is monitoring the average request duration during each 300 sec
interval for requests processed by a web proxy during the interval.
In this case, the proxy exposes two counters, the number of requests
processed since boot and the total cumulative duration of all processed
requests. Clearly these counters both have some rollover point, but using the
DERIVE data source also handles the reset that occurs when the web proxy is
stopped and restarted.

In the B<RRD>, the first data source stores the requests per second rate
during the interval. The second data source stores the total duration of all
requests processed during the interval divided by 300. The COMPUTE data source
divides each PDP of the AccumDuration by the corresponding PDP of
TotalRequests and stores the average request duration. The remainder of the
RPN expression handles the divide by zero case.

=head1 AUTHOR

Tobias Oetiker E<lt>tobi@oetiker.chE<gt>