Code

Imported upstream version 1.4.3.
[pkg-rrdtool.git] / doc / rrdcreate.html
1 <?xml version="1.0" ?>
2 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">
3 <html xmlns="http://www.w3.org/1999/xhtml">
4 <head>
5 <title>rrdcreate</title>
6 <meta http-equiv="content-type" content="text/html; charset=utf-8" />
7 <link rev="made" href="mailto:root@localhost" />
8 </head>
10 <body style="background-color: white">
13 <!-- INDEX BEGIN -->
14 <div name="index">
15 <p><a name="__index__"></a></p>
16 <!--
18 <ul>
20         <li><a href="#name">NAME</a></li>
21         <li><a href="#synopsis">SYNOPSIS</a></li>
22         <li><a href="#description">DESCRIPTION</a></li>
23         <ul>
25                 <li><a href="#filename"><em>filename</em></a></li>
26                 <li><a href="#__start__b_start_time__default__now___10s_"><strong>-start</strong>|<strong>-b</strong> <em>start time</em> (default: now - 10s)</a></li>
27                 <li><a href="#__step__s_step__default__300_seconds_"><strong>-step</strong>|<strong>-s</strong> <em>step</em> (default: 300 seconds)</a></li>
28                 <li><a href="#__no_overwrite"><strong>-no-overwrite</strong></a></li>
29                 <li><a href="#ds_ds_name_dst_dst_arguments"><strong>DS:</strong><em>ds-name</em><strong>:</strong><em>DST</em><strong>:</strong><em>dst arguments</em></a></li>
30                 <li><a href="#rra_cf_cf_arguments"><strong>RRA:</strong><em>CF</em><strong>:</strong><em>cf arguments</em></a></li>
31         </ul>
33         <li><a href="#aberrant_behavior_detection_with_holt_winters_forecasting">Aberrant Behavior Detection with Holt-Winters Forecasting</a></li>
34         <li><a href="#the_heartbeat_and_the_step">The HEARTBEAT and the STEP</a></li>
35         <li><a href="#how_to_measure">HOW TO MEASURE</a></li>
36         <li><a href="#example">EXAMPLE</a></li>
37         <li><a href="#example_2">EXAMPLE 2</a></li>
38         <li><a href="#example_3">EXAMPLE 3</a></li>
39         <li><a href="#author">AUTHOR</a></li>
40 </ul>
42 -->
45 </div>
46 <!-- INDEX END -->
48 <p>
49 </p>
50 <h1><a name="name">NAME</a></h1>
51 <p>rrdcreate - Set up a new Round Robin Database</p>
52 <p>
53 </p>
54 <hr />
55 <h1><a name="synopsis">SYNOPSIS</a></h1>
56 <p><strong>rrdtool</strong> <strong>create</strong> <em>filename</em>
57 [<strong>--start</strong>|<strong>-b</strong>&nbsp;<em>start&nbsp;time</em>]
58 [<strong>--step</strong>|<strong>-s</strong>&nbsp;<em>step</em>]
59 [<strong>--no-overwrite</strong>]
60 [<strong>DS:</strong><em>ds-name</em><strong>:</strong><em>DST</em><strong>:</strong><em>dst&nbsp;arguments</em>]
61 [<strong>RRA:</strong><em>CF</em><strong>:</strong><em>cf&nbsp;arguments</em>]</p>
62 <p>
63 </p>
64 <hr />
65 <h1><a name="description">DESCRIPTION</a></h1>
66 <p>The create function of RRDtool lets you set up new Round Robin
67 Database (<strong>RRD</strong>) files.  The file is created at its final, full size
68 and filled with <em>*UNKNOWN*</em> data.</p>
69 <p>
70 </p>
71 <h2><a name="filename"><em>filename</em></a></h2>
72 <p>The name of the <strong>RRD</strong> you want to create. <strong>RRD</strong> files should end
73 with the extension <em class="file">.rrd</em>. However, <strong>RRDtool</strong> will accept any
74 filename.</p>
75 <p>
76 </p>
77 <h2><a name="__start__b_start_time__default__now___10s_"><strong>--start</strong>|<strong>-b</strong> <em>start time</em> (default: now - 10s)</a></h2>
78 <p>Specifies the time in seconds since 1970-01-01 UTC when the first
79 value should be added to the <strong>RRD</strong>. <strong>RRDtool</strong> will not accept
80 any data timed before or at the time specified.</p>
81 <p>See also AT-STYLE TIME SPECIFICATION section in the
82 <em>rrdfetch</em> documentation for other ways to specify time.</p>
83 <p>
84 </p>
85 <h2><a name="__step__s_step__default__300_seconds_"><strong>--step</strong>|<strong>-s</strong> <em>step</em> (default: 300 seconds)</a></h2>
86 <p>Specifies the base interval in seconds with which data will be fed
87 into the <strong>RRD</strong>.</p>
88 <p>
89 </p>
90 <h2><a name="__no_overwrite"><strong>--no-overwrite</strong></a></h2>
91 <p>Do not clobber an existing file of the same name.</p>
92 <p>
93 </p>
94 <h2><a name="ds_ds_name_dst_dst_arguments"><strong>DS:</strong><em>ds-name</em><strong>:</strong><em>DST</em><strong>:</strong><em>dst arguments</em></a></h2>
95 <p>A single <strong>RRD</strong> can accept input from several data sources (<strong>DS</strong>),
96 for example incoming and outgoing traffic on a specific communication
97 line. With the <strong>DS</strong> configuration option you must define some basic
98 properties of each data source you want to store in the <strong>RRD</strong>.</p>
99 <p><em>ds-name</em> is the name you will use to reference this particular data
100 source from an <strong>RRD</strong>. A <em>ds-name</em> must be 1 to 19 characters long in
101 the characters [a-zA-Z0-9_].</p>
102 <p><em>DST</em> defines the Data Source Type. The remaining arguments of a
103 data source entry depend on the data source type. For GAUGE, COUNTER,
104 DERIVE, and ABSOLUTE the format for a data source entry is:</p>
105 <p><strong>DS:</strong><em>ds-name</em><strong>:</strong><em>GAUGE | COUNTER | DERIVE | ABSOLUTE</em><strong>:</strong><em>heartbeat</em><strong>:</strong><em>min</em><strong>:</strong><em>max</em></p>
106 <p>For COMPUTE data sources, the format is:</p>
107 <p><strong>DS:</strong><em>ds-name</em><strong>:</strong><em>COMPUTE</em><strong>:</strong><em>rpn-expression</em></p>
108 <p>In order to decide which data source type to use, review the
109 definitions that follow. Also consult the section on &quot;HOW TO MEASURE&quot;
110 for further insight.</p>
111 <dl>
112 <dt><strong><a name="gauge" class="item"><strong>GAUGE</strong></a></strong></dt>
114 <dd>
115 <p>is for things like temperatures or number of people in a room or the
116 value of a RedHat share.</p>
117 </dd>
118 <dt><strong><a name="counter" class="item"><strong>COUNTER</strong></a></strong></dt>
120 <dd>
121 <p>is for continuous incrementing counters like the ifInOctets counter in
122 a router. The <strong>COUNTER</strong> data source assumes that the counter never
123 decreases, except when a counter overflows.  The update function takes
124 the overflow into account.  The counter is stored as a per-second
125 rate. When the counter overflows, RRDtool checks if the overflow
126 happened at the 32bit or 64bit border and acts accordingly by adding
127 an appropriate value to the result.</p>
128 </dd>
129 <dt><strong><a name="derive" class="item"><strong>DERIVE</strong></a></strong></dt>
131 <dd>
132 <p>will store the derivative of the line going from the last to the
133 current value of the data source. This can be useful for gauges, for
134 example, to measure the rate of people entering or leaving a
135 room. Internally, derive works exactly like COUNTER but without
136 overflow checks. So if your counter does not reset at 32 or 64 bit you
137 might want to use DERIVE and combine it with a MIN value of 0.</p>
138 <p><strong>NOTE on COUNTER vs DERIVE</strong></p>
139 <p>by Don Baarda &lt;<a href="mailto:don.baarda@baesystems.com">don.baarda@baesystems.com</a>&gt;</p>
140 <p>If you cannot tolerate ever mistaking the occasional counter reset for a
141 legitimate counter wrap, and would prefer &quot;Unknowns&quot; for all legitimate
142 counter wraps and resets, always use DERIVE with min=0. Otherwise, using
143 COUNTER with a suitable max will return correct values for all legitimate
144 counter wraps, mark some counter resets as &quot;Unknown&quot;, but can mistake some
145 counter resets for a legitimate counter wrap.</p>
146 <p>For a 5 minute step and 32-bit counter, the probability of mistaking a
147 counter reset for a legitimate wrap is arguably about 0.8% per 1Mbps of
148 maximum bandwidth. Note that this equates to 80% for 100Mbps interfaces, so
149 for high bandwidth interfaces and a 32bit counter, DERIVE with min=0 is
150 probably preferable. If you are using a 64bit counter, just about any max
151 setting will eliminate the possibility of mistaking a reset for a counter
152 wrap.</p>
153 </dd>
154 <dt><strong><a name="absolute" class="item"><strong>ABSOLUTE</strong></a></strong></dt>
156 <dd>
157 <p>is for counters which get reset upon reading. This is used for fast counters
158 which tend to overflow. So instead of reading them normally you reset them
159 after every read to make sure you have a maximum time available before the
160 next overflow. Another usage is for things you count like number of messages
161 since the last update.</p>
162 </dd>
163 <dt><strong><a name="compute" class="item"><strong>COMPUTE</strong></a></strong></dt>
165 <dd>
166 <p>is for storing the result of a formula applied to other data sources
167 in the <strong>RRD</strong>. This data source is not supplied a value on update, but
168 rather its Primary Data Points (PDPs) are computed from the PDPs of
169 the data sources according to the rpn-expression that defines the
170 formula. Consolidation functions are then applied normally to the PDPs
171 of the COMPUTE data source (that is the rpn-expression is only applied
172 to generate PDPs). In database software, such data sets are referred
173 to as &quot;virtual&quot; or &quot;computed&quot; columns.</p>
174 </dd>
175 </dl>
176 <p><em>heartbeat</em> defines the maximum number of seconds that may pass
177 between two updates of this data source before the value of the
178 data source is assumed to be <em>*UNKNOWN*</em>.</p>
179 <p><em>min</em> and <em>max</em> define the expected range values for data supplied by a
180 data source. If <em>min</em> and/or <em>max</em> any value outside the defined range
181 will be regarded as <em>*UNKNOWN*</em>. If you do not know or care about min and
182 max, set them to U for unknown. Note that min and max always refer to the
183 processed values of the DS. For a traffic-<strong>COUNTER</strong> type DS this would be
184 the maximum and minimum data-rate expected from the device.</p>
185 <p><em>If information on minimal/maximal expected values is available,
186 always set the min and/or max properties. This will help RRDtool in
187 doing a simple sanity check on the data supplied when running update.</em></p>
188 <p><em>rpn-expression</em> defines the formula used to compute the PDPs of a
189 COMPUTE data source from other data sources in the same &lt;RRD&gt;. It is
190 similar to defining a <strong>CDEF</strong> argument for the graph command. Please
191 refer to that manual page for a list and description of RPN operations
192 supported. For COMPUTE data sources, the following RPN operations are
193 not supported: COUNT, PREV, TIME, and LTIME. In addition, in defining
194 the RPN expression, the COMPUTE data source may only refer to the
195 names of data source listed previously in the create command. This is
196 similar to the restriction that <strong>CDEF</strong>s must refer only to <strong>DEF</strong>s
197 and <strong>CDEF</strong>s previously defined in the same graph command.</p>
198 <p>
199 </p>
200 <h2><a name="rra_cf_cf_arguments"><strong>RRA:</strong><em>CF</em><strong>:</strong><em>cf arguments</em></a></h2>
201 <p>The purpose of an <strong>RRD</strong> is to store data in the round robin archives
202 (<strong>RRA</strong>). An archive consists of a number of data values or statistics for
203 each of the defined data-sources (<strong>DS</strong>) and is defined with an <strong>RRA</strong> line.</p>
204 <p>When data is entered into an <strong>RRD</strong>, it is first fit into time slots
205 of the length defined with the <strong>-s</strong> option, thus becoming a <em>primary
206 data point</em>.</p>
207 <p>The data is also processed with the consolidation function (<em>CF</em>) of
208 the archive. There are several consolidation functions that
209 consolidate primary data points via an aggregate function: <strong>AVERAGE</strong>,
210 <strong>MIN</strong>, <strong>MAX</strong>, <strong>LAST</strong>.</p>
211 <dl>
212 <dt><strong><a name="average" class="item">AVERAGE</a></strong></dt>
214 <dd>
215 <p>the average of the data points is stored.</p>
216 </dd>
217 <dt><strong><a name="min" class="item">MIN</a></strong></dt>
219 <dd>
220 <p>the smallest of the data points is stored.</p>
221 </dd>
222 <dt><strong><a name="max" class="item">MAX</a></strong></dt>
224 <dd>
225 <p>the largest of the data points is stored.</p>
226 </dd>
227 <dt><strong><a name="last" class="item">LAST</a></strong></dt>
229 <dd>
230 <p>the last data points is used.</p>
231 </dd>
232 </dl>
233 <p>Note that data aggregation inevitably leads to loss of precision and
234 information. The trick is to pick the aggregate function such that the
235 <em>interesting</em> properties of your data is kept across the aggregation
236 process.</p>
237 <p>The format of <strong>RRA</strong> line for these
238 consolidation functions is:</p>
239 <p><strong>RRA:</strong><em>AVERAGE | MIN | MAX | LAST</em><strong>:</strong><em>xff</em><strong>:</strong><em>steps</em><strong>:</strong><em>rows</em></p>
240 <p><em>xff</em> The xfiles factor defines what part of a consolidation interval may
241 be made up from <em>*UNKNOWN*</em> data while the consolidated value is still
242 regarded as known. It is given as the ratio of allowed <em>*UNKNOWN*</em> PDPs
243 to the number of PDPs in the interval. Thus, it ranges from 0 to 1 (exclusive).</p>
244 <p><em>steps</em> defines how many of these <em>primary data points</em> are used to build
245 a <em>consolidated data point</em> which then goes into the archive.</p>
246 <p><em>rows</em> defines how many generations of data values are kept in an <strong>RRA</strong>.
247 Obviously, this has to be greater than zero.</p>
248 <p>
249 </p>
250 <hr />
251 <h1><a name="aberrant_behavior_detection_with_holt_winters_forecasting">Aberrant Behavior Detection with Holt-Winters Forecasting</a></h1>
252 <p>In addition to the aggregate functions, there are a set of specialized
253 functions that enable <strong>RRDtool</strong> to provide data smoothing (via the
254 Holt-Winters forecasting algorithm), confidence bands, and the
255 flagging aberrant behavior in the data source time series:</p>
256 <ul>
257 <li>
258 <p><strong>RRA:</strong><em>HWPREDICT</em><strong>:</strong><em>rows</em><strong>:</strong><em>alpha</em><strong>:</strong><em>beta</em><strong>:</strong><em>seasonal period</em>[<strong>:</strong><em>rra-num</em>]</p>
259 </li>
260 <li>
261 <p><strong>RRA:</strong><em>MHWPREDICT</em><strong>:</strong><em>rows</em><strong>:</strong><em>alpha</em><strong>:</strong><em>beta</em><strong>:</strong><em>seasonal period</em>[<strong>:</strong><em>rra-num</em>]</p>
262 </li>
263 <li>
264 <p><strong>RRA:</strong><em>SEASONAL</em><strong>:</strong><em>seasonal period</em><strong>:</strong><em>gamma</em><strong>:</strong><em>rra-num</em>[<strong>:smoothing-window=</strong><em>fraction</em>]</p>
265 </li>
266 <li>
267 <p><strong>RRA:</strong><em>DEVSEASONAL</em><strong>:</strong><em>seasonal period</em><strong>:</strong><em>gamma</em><strong>:</strong><em>rra-num</em>[<strong>:smoothing-window=</strong><em>fraction</em>]</p>
268 </li>
269 <li>
270 <p><strong>RRA:</strong><em>DEVPREDICT</em><strong>:</strong><em>rows</em><strong>:</strong><em>rra-num</em></p>
271 </li>
272 <li>
273 <p><strong>RRA:</strong><em>FAILURES</em><strong>:</strong><em>rows</em><strong>:</strong><em>threshold</em><strong>:</strong><em>window length</em><strong>:</strong><em>rra-num</em></p>
274 </li>
275 </ul>
276 <p>These <strong>RRAs</strong> differ from the true consolidation functions in several ways.
277 First, each of the <strong>RRA</strong>s is updated once for every primary data point.
278 Second, these <strong>RRAs</strong> are interdependent. To generate real-time confidence
279 bounds, a matched set of SEASONAL, DEVSEASONAL, DEVPREDICT, and either
280 HWPREDICT or MHWPREDICT must exist. Generating smoothed values of the primary
281 data points requires a SEASONAL <strong>RRA</strong> and either an HWPREDICT or MHWPREDICT 
282 <strong>RRA</strong>. Aberrant behavior detection requires FAILURES, DEVSEASONAL, SEASONAL,
283 and either HWPREDICT or MHWPREDICT.</p>
284 <p>The predicted, or smoothed, values are stored in the HWPREDICT or MHWPREDICT
285 <strong>RRA</strong>. HWPREDICT and MHWPREDICT are actually two variations on the
286 Holt-Winters method. They are interchangeable. Both attempt to decompose data
287 into three components: a baseline, a trend, and a seasonal coefficient.
288 HWPREDICT adds its seasonal coefficient to the baseline to form a prediction, whereas
289 MHWPREDICT multiplies its seasonal coefficient by the baseline to form a
290 prediction. The difference is noticeable when the baseline changes
291 significantly in the course of a season; HWPREDICT will predict the seasonality
292 to stay constant as the baseline changes, but MHWPREDICT will predict the
293 seasonality to grow or shrink in proportion to the baseline. The proper choice
294 of method depends on the thing being modeled. For simplicity, the rest of this
295 discussion will refer to HWPREDICT, but MHWPREDICT may be substituted in its
296 place.</p>
297 <p>The predicted deviations are stored in DEVPREDICT (think a standard deviation
298 which can be scaled to yield a confidence band). The FAILURES <strong>RRA</strong> stores 
299 binary indicators. A 1 marks the indexed observation as failure; that is, the 
300 number of confidence bounds violations in the preceding window of observations 
301 met or exceeded a specified threshold. An example of using these <strong>RRAs</strong> to graph 
302 confidence bounds and failures appears in <a href="././rrdgraph.html">the rrdgraph manpage</a>.</p>
303 <p>The SEASONAL and DEVSEASONAL <strong>RRAs</strong> store the seasonal coefficients for the
304 Holt-Winters forecasting algorithm and the seasonal deviations, respectively.
305 There is one entry per observation time point in the seasonal cycle. For
306 example, if primary data points are generated every five minutes and the
307 seasonal cycle is 1 day, both SEASONAL and DEVSEASONAL will have 288 rows.</p>
308 <p>In order to simplify the creation for the novice user, in addition to
309 supporting explicit creation of the HWPREDICT, SEASONAL, DEVPREDICT,
310 DEVSEASONAL, and FAILURES <strong>RRAs</strong>, the <strong>RRDtool</strong> create command supports
311 implicit creation of the other four when HWPREDICT is specified alone and
312 the final argument <em>rra-num</em> is omitted.</p>
313 <p><em>rows</em> specifies the length of the <strong>RRA</strong> prior to wrap around. Remember
314 that there is a one-to-one correspondence between primary data points and
315 entries in these RRAs. For the HWPREDICT CF, <em>rows</em> should be larger than
316 the <em>seasonal period</em>. If the DEVPREDICT <strong>RRA</strong> is implicitly created, the
317 default number of rows is the same as the HWPREDICT <em>rows</em> argument. If the
318 FAILURES <strong>RRA</strong> is implicitly created, <em>rows</em> will be set to the <em>seasonal
319 period</em> argument of the HWPREDICT <strong>RRA</strong>. Of course, the <strong>RRDtool</strong>
320 <em>resize</em> command is available if these defaults are not sufficient and the
321 creator wishes to avoid explicit creations of the other specialized function
322 <strong>RRAs</strong>.</p>
323 <p><em>seasonal period</em> specifies the number of primary data points in a seasonal
324 cycle. If SEASONAL and DEVSEASONAL are implicitly created, this argument for
325 those <strong>RRAs</strong> is set automatically to the value specified by HWPREDICT. If
326 they are explicitly created, the creator should verify that all three
327 <em>seasonal period</em> arguments agree.</p>
328 <p><em>alpha</em> is the adaption parameter of the intercept (or baseline)
329 coefficient in the Holt-Winters forecasting algorithm. See <a href="././rrdtool.html">the rrdtool manpage</a> for a
330 description of this algorithm. <em>alpha</em> must lie between 0 and 1. A value
331 closer to 1 means that more recent observations carry greater weight in
332 predicting the baseline component of the forecast. A value closer to 0 means
333 that past history carries greater weight in predicting the baseline
334 component.</p>
335 <p><em>beta</em> is the adaption parameter of the slope (or linear trend) coefficient
336 in the Holt-Winters forecasting algorithm. <em>beta</em> must lie between 0 and 1
337 and plays the same role as <em>alpha</em> with respect to the predicted linear
338 trend.</p>
339 <p><em>gamma</em> is the adaption parameter of the seasonal coefficients in the
340 Holt-Winters forecasting algorithm (HWPREDICT) or the adaption parameter in
341 the exponential smoothing update of the seasonal deviations. It must lie
342 between 0 and 1. If the SEASONAL and DEVSEASONAL <strong>RRAs</strong> are created
343 implicitly, they will both have the same value for <em>gamma</em>: the value
344 specified for the HWPREDICT <em>alpha</em> argument. Note that because there is
345 one seasonal coefficient (or deviation) for each time point during the
346 seasonal cycle, the adaptation rate is much slower than the baseline. Each
347 seasonal coefficient is only updated (or adapts) when the observed value
348 occurs at the offset in the seasonal cycle corresponding to that
349 coefficient.</p>
350 <p>If SEASONAL and DEVSEASONAL <strong>RRAs</strong> are created explicitly, <em>gamma</em> need not
351 be the same for both. Note that <em>gamma</em> can also be changed via the
352 <strong>RRDtool</strong> <em>tune</em> command.</p>
353 <p><em>smoothing-window</em> specifies the fraction of a season that should be
354 averaged around each point. By default, the value of <em>smoothing-window</em> is
355 0.05, which means each value in SEASONAL and DEVSEASONAL will be occasionally
356 replaced by averaging it with its (<em>seasonal period</em>*0.05) nearest neighbors.
357 Setting <em>smoothing-window</em> to zero will disable the running-average smoother
358 altogether.</p>
359 <p><em>rra-num</em> provides the links between related <strong>RRAs</strong>. If HWPREDICT is
360 specified alone and the other <strong>RRAs</strong> are created implicitly, then
361 there is no need to worry about this argument. If <strong>RRAs</strong> are created
362 explicitly, then carefully pay attention to this argument. For each
363 <strong>RRA</strong> which includes this argument, there is a dependency between
364 that <strong>RRA</strong> and another <strong>RRA</strong>. The <em>rra-num</em> argument is the 1-based
365 index in the order of <strong>RRA</strong> creation (that is, the order they appear
366 in the <em>create</em> command). The dependent <strong>RRA</strong> for each <strong>RRA</strong>
367 requiring the <em>rra-num</em> argument is listed here:</p>
368 <ul>
369 <li>
370 <p>HWPREDICT <em>rra-num</em> is the index of the SEASONAL <strong>RRA</strong>.</p>
371 </li>
372 <li>
373 <p>SEASONAL <em>rra-num</em> is the index of the HWPREDICT <strong>RRA</strong>.</p>
374 </li>
375 <li>
376 <p>DEVPREDICT <em>rra-num</em> is the index of the DEVSEASONAL <strong>RRA</strong>.</p>
377 </li>
378 <li>
379 <p>DEVSEASONAL <em>rra-num</em> is the index of the HWPREDICT <strong>RRA</strong>.</p>
380 </li>
381 <li>
382 <p>FAILURES <em>rra-num</em> is the index of the DEVSEASONAL <strong>RRA</strong>.</p>
383 </li>
384 </ul>
385 <p><em>threshold</em> is the minimum number of violations (observed values outside
386 the confidence bounds) within a window that constitutes a failure. If the
387 FAILURES <strong>RRA</strong> is implicitly created, the default value is 7.</p>
388 <p><em>window length</em> is the number of time points in the window. Specify an
389 integer greater than or equal to the threshold and less than or equal to 28.
390 The time interval this window represents depends on the interval between
391 primary data points. If the FAILURES <strong>RRA</strong> is implicitly created, the
392 default value is 9.</p>
393 <p>
394 </p>
395 <hr />
396 <h1><a name="the_heartbeat_and_the_step">The HEARTBEAT and the STEP</a></h1>
397 <p>Here is an explanation by Don Baarda on the inner workings of RRDtool.
398 It may help you to sort out why all this *UNKNOWN* data is popping
399 up in your databases:</p>
400 <p>RRDtool gets fed samples/updates at arbitrary times. From these it builds Primary
401 Data Points (PDPs) on every &quot;step&quot; interval. The PDPs are
402 then accumulated into the RRAs.</p>
403 <p>The &quot;heartbeat&quot; defines the maximum acceptable interval between
404 samples/updates. If the interval between samples is less than &quot;heartbeat&quot;,
405 then an average rate is calculated and applied for that interval. If
406 the interval between samples is longer than &quot;heartbeat&quot;, then that
407 entire interval is considered &quot;unknown&quot;. Note that there are other
408 things that can make a sample interval &quot;unknown&quot;, such as the rate
409 exceeding limits, or a sample that was explicitly marked as unknown.</p>
410 <p>The known rates during a PDP's &quot;step&quot; interval are used to calculate
411 an average rate for that PDP. If the total &quot;unknown&quot; time accounts for
412 more than <strong>half</strong> the &quot;step&quot;, the entire PDP is marked
413 as &quot;unknown&quot;. This means that a mixture of known and &quot;unknown&quot; sample
414 times in a single PDP &quot;step&quot; may or may not add up to enough &quot;known&quot;
415 time to warrant a known PDP.</p>
416 <p>The &quot;heartbeat&quot; can be short (unusual) or long (typical) relative to
417 the &quot;step&quot; interval between PDPs. A short &quot;heartbeat&quot; means you
418 require multiple samples per PDP, and if you don't get them mark the
419 PDP unknown. A long heartbeat can span multiple &quot;steps&quot;, which means
420 it is acceptable to have multiple PDPs calculated from a single
421 sample. An extreme example of this might be a &quot;step&quot; of 5 minutes and a
422 &quot;heartbeat&quot; of one day, in which case a single sample every day will
423 result in all the PDPs for that entire day period being set to the
424 same average rate. <em>-- Don Baarda &lt;<a href="mailto:don.baarda@baesystems.com">don.baarda@baesystems.com</a>&gt;</em></p>
425 <pre>
426        time|
427        axis|
428  begin__|00|
429         |01|
430        u|02|----* sample1, restart &quot;hb&quot;-timer
431        u|03|   /
432        u|04|  /
433        u|05| /
434        u|06|/     &quot;hbt&quot; expired
435        u|07|
436         |08|----* sample2, restart &quot;hb&quot; 
437         |09|   / 
438         |10|  /
439        u|11|----* sample3, restart &quot;hb&quot;
440        u|12|   /
441        u|13|  /
442  step1_u|14| /
443        u|15|/     &quot;swt&quot; expired
444        u|16|
445         |17|----* sample4, restart &quot;hb&quot;, create &quot;pdp&quot; for step1 = 
446         |18|   /  = unknown due to 10 &quot;u&quot; labled secs &gt; 0.5 * step
447         |19|  /
448         |20| /
449         |21|----* sample5, restart &quot;hb&quot;
450         |22|   /
451         |23|  /
452         |24|----* sample6, restart &quot;hb&quot;
453         |25|   /
454         |26|  /
455         |27|----* sample7, restart &quot;hb&quot;
456  step2__|28|   /
457         |22|  /
458         |23|----* sample8, restart &quot;hb&quot;, create &quot;pdp&quot; for step1, create &quot;cdp&quot; 
459         |24|   /
460         |25|  /</pre>
461 <p>graphics by <em><a href="mailto:vladimir.lavrov@desy.de">vladimir.lavrov@desy.de</a></em>.</p>
462 <p>
463 </p>
464 <hr />
465 <h1><a name="how_to_measure">HOW TO MEASURE</a></h1>
466 <p>Here are a few hints on how to measure:</p>
467 <dl>
468 <dt><strong><a name="temperature" class="item">Temperature</a></strong></dt>
470 <dd>
471 <p>Usually you have some type of meter you can read to get the temperature.
472 The temperature is not really connected with a time. The only connection is
473 that the temperature reading happened at a certain time. You can use the
474 <strong>GAUGE</strong> data source type for this. RRDtool will then record your reading
475 together with the time.</p>
476 </dd>
477 <dt><strong><a name="mail_messages" class="item">Mail Messages</a></strong></dt>
479 <dd>
480 <p>Assume you have a method to count the number of messages transported by
481 your mail server in a certain amount of time, giving you data like '5
482 messages in the last 65 seconds'. If you look at the count of 5 like an
483 <strong>ABSOLUTE</strong> data type you can simply update the RRD with the number 5 and the
484 end time of your monitoring period. RRDtool will then record the number of
485 messages per second. If at some later stage you want to know the number of
486 messages transported in a day, you can get the average messages per second
487 from RRDtool for the day in question and multiply this number with the
488 number of seconds in a day. Because all math is run with Doubles, the
489 precision should be acceptable.</p>
490 </dd>
491 <dt><strong><a name="it_s_always_a_rate" class="item">It's always a Rate</a></strong></dt>
493 <dd>
494 <p>RRDtool stores rates in amount/second for COUNTER, DERIVE and ABSOLUTE
495 data.  When you plot the data, you will get on the y axis
496 amount/second which you might be tempted to convert to an absolute
497 amount by multiplying by the delta-time between the points. RRDtool
498 plots continuous data, and as such is not appropriate for plotting
499 absolute amounts as for example &quot;total bytes&quot; sent and received in a
500 router. What you probably want is plot rates that you can scale to
501 bytes/hour, for example, or plot absolute amounts with another tool
502 that draws bar-plots, where the delta-time is clear on the plot for
503 each point (such that when you read the graph you see for example GB
504 on the y axis, days on the x axis and one bar for each day).</p>
505 </dd>
506 </dl>
507 <p>
508 </p>
509 <hr />
510 <h1><a name="example">EXAMPLE</a></h1>
511 <pre>
512  rrdtool create temperature.rrd --step 300 \
513   DS:temp:GAUGE:600:-273:5000 \
514   RRA:AVERAGE:0.5:1:1200 \
515   RRA:MIN:0.5:12:2400 \
516   RRA:MAX:0.5:12:2400 \
517   RRA:AVERAGE:0.5:12:2400</pre>
518 <p>This sets up an <strong>RRD</strong> called <em class="file">temperature.rrd</em> which accepts one
519 temperature value every 300 seconds. If no new data is supplied for
520 more than 600 seconds, the temperature becomes <em>*UNKNOWN*</em>.  The
521 minimum acceptable value is -273 and the maximum is 5'000.</p>
522 <p>A few archive areas are also defined. The first stores the
523 temperatures supplied for 100 hours (1'200 * 300 seconds = 100
524 hours). The second RRA stores the minimum temperature recorded over
525 every hour (12 * 300 seconds = 1 hour), for 100 days (2'400 hours). The
526 third and the fourth RRA's do the same for the maximum and
527 average temperature, respectively.</p>
528 <p>
529 </p>
530 <hr />
531 <h1><a name="example_2">EXAMPLE 2</a></h1>
532 <pre>
533  rrdtool create monitor.rrd --step 300        \
534    DS:ifOutOctets:COUNTER:1800:0:4294967295   \
535    RRA:AVERAGE:0.5:1:2016                     \
536    RRA:HWPREDICT:1440:0.1:0.0035:288</pre>
537 <p>This example is a monitor of a router interface. The first <strong>RRA</strong> tracks the
538 traffic flow in octets; the second <strong>RRA</strong> generates the specialized
539 functions <strong>RRAs</strong> for aberrant behavior detection. Note that the <em>rra-num</em>
540 argument of HWPREDICT is missing, so the other <strong>RRAs</strong> will implicitly be
541 created with default parameter values. In this example, the forecasting
542 algorithm baseline adapts quickly; in fact the most recent one hour of
543 observations (each at 5 minute intervals) accounts for 75% of the baseline
544 prediction. The linear trend forecast adapts much more slowly. Observations
545 made during the last day (at 288 observations per day) account for only
546 65% of the predicted linear trend. Note: these computations rely on an
547 exponential smoothing formula described in the LISA 2000 paper.</p>
548 <p>The seasonal cycle is one day (288 data points at 300 second intervals), and
549 the seasonal adaption parameter will be set to 0.1. The RRD file will store 5
550 days (1'440 data points) of forecasts and deviation predictions before wrap
551 around. The file will store 1 day (a seasonal cycle) of 0-1 indicators in
552 the FAILURES <strong>RRA</strong>.</p>
553 <p>The same RRD file and <strong>RRAs</strong> are created with the following command,
554 which explicitly creates all specialized function <strong>RRAs</strong>.</p>
555 <pre>
556  rrdtool create monitor.rrd --step 300 \
557    DS:ifOutOctets:COUNTER:1800:0:4294967295 \
558    RRA:AVERAGE:0.5:1:2016 \
559    RRA:HWPREDICT:1440:0.1:0.0035:288:3 \
560    RRA:SEASONAL:288:0.1:2 \
561    RRA:DEVPREDICT:1440:5 \
562    RRA:DEVSEASONAL:288:0.1:2 \
563    RRA:FAILURES:288:7:9:5</pre>
564 <p>Of course, explicit creation need not replicate implicit create, a
565 number of arguments could be changed.</p>
566 <p>
567 </p>
568 <hr />
569 <h1><a name="example_3">EXAMPLE 3</a></h1>
570 <pre>
571  rrdtool create proxy.rrd --step 300 \
572    DS:Total:DERIVE:1800:0:U  \
573    DS:Duration:DERIVE:1800:0:U  \
574    DS:AvgReqDur:COMPUTE:Duration,Requests,0,EQ,1,Requests,IF,/ \
575    RRA:AVERAGE:0.5:1:2016</pre>
576 <p>This example is monitoring the average request duration during each 300 sec
577 interval for requests processed by a web proxy during the interval.
578 In this case, the proxy exposes two counters, the number of requests
579 processed since boot and the total cumulative duration of all processed
580 requests. Clearly these counters both have some rollover point, but using the
581 DERIVE data source also handles the reset that occurs when the web proxy is
582 stopped and restarted.</p>
583 <p>In the <strong>RRD</strong>, the first data source stores the requests per second rate
584 during the interval. The second data source stores the total duration of all
585 requests processed during the interval divided by 300. The COMPUTE data source
586 divides each PDP of the AccumDuration by the corresponding PDP of
587 TotalRequests and stores the average request duration. The remainder of the
588 RPN expression handles the divide by zero case.</p>
589 <p>
590 </p>
591 <hr />
592 <h1><a name="author">AUTHOR</a></h1>
593 <p>Tobias Oetiker &lt;<a href="mailto:tobi@oetiker.ch">tobi@oetiker.ch</a>&gt;</p>
595 </body>
597 </html>