org.jrobin.data

Class DataProcessor

public class DataProcessor extends Object implements ConsolFuns

Class which should be used for all calculations based on the data fetched from RRD files. This class supports ordinary DEF datasources (defined in RRD files), CDEF datasources (RPN expressions evaluation), SDEF (static datasources - extension of JRobin) and PDEF (plottables, see Plottable for more information.

Typical class usage:

 final long t1 = ...
 final long t2 = ...
 DataProcessor dp = new DataProcessor(t1, t2);
 // DEF datasource
 dp.addDatasource("x", "demo.rrd", "some_source", "AVERAGE");
 // DEF datasource
 dp.addDatasource("y", "demo.rrd", "some_other_source", "AVERAGE");
 // CDEF datasource, z = (x + y) / 2
 dp.addDatasource("z", "x,y,+,2,/");
 // ACTION!
 dp.processData();
 // Dump calculated values
 System.out.println(dp.dump());
 
Field Summary
static intDEFAULT_PIXEL_COUNT
Constant representing the default number of pixels on a JRobin graph (will be used if no other value is specified with setStep() method.
static booleanDEFAULT_POOL_USAGE_POLICY
Constant that defines the default RrdDbPool usage policy.
Constructor Summary
DataProcessor(long t1, long t2)
Creates new DataProcessor object for the given time span.
DataProcessor(Date d1, Date d2)
Creates new DataProcessor object for the given time span.
DataProcessor(Calendar gc1, Calendar gc2)
Creates new DataProcessor object for the given time span.
Method Summary
voidaddDatasource(String name, Plottable plottable)

Adds a custom, plottable datasource (PDEF).

voidaddDatasource(String name, String rpnExpression)

Adds complex source (CDEF).

voidaddDatasource(String name, String defName, String consolFun)

Adds static source (SDEF).

voidaddDatasource(String name, String file, String dsName, String consolFunc)

Adds simple datasource (DEF).

voidaddDatasource(String name, String file, String dsName, String consolFunc, String backend)

Adds simple source (DEF).

voidaddDatasource(String name, FetchData fetchData)
Adds DEF datasource with datasource values already available in the FetchData object.
Stringdump()
Dumps timestamps and values of all datasources in a tabelar form.
doubleget95Percentile(String sourceName)
This method is just an alias for getPercentile method.
doublegetAggregate(String sourceName, String consolFun)
Returns single aggregated value for a single datasource.
AggregatesgetAggregates(String sourceName)
Returns all (MIN, MAX, LAST, FIRST, AVERAGE and TOTAL) aggregated values for a single datasource.
longgetEndingTimestamp()
Returns ending timestamp.
longgetFetchRequestResolution()
Returns desired RRD archive step (reslution) in seconds to be used while fetching data from RRD files.
longgetLastRrdArchiveUpdateTime()
Returns time when last RRD archive was updated (all RRD files are considered).
doublegetPercentile(String sourceName)
Used by ISPs which charge for bandwidth utilization on a "95th percentile" basis.

The 95th percentile is the highest source value left when the top 5% of a numerically sorted set of source data is discarded.

doublegetPercentile(String sourceName, double percentile)
The same as getPercentile but with a possibility to define custom percentile boundary (different from 95).
intgetPixelCount()
Returns the number of pixels (target graph width).
String[]getSourceNames()
Returns array of datasource names defined in this DataProcessor.
longgetStep()
Returns the time step used for data processing.
long[]getTimestamps()
Returns consolidated timestamps created with the processData method.
long[]getTimestampsPerPixel(int pixelCount)
Calculates timestamps which correspond to individual pixels on the graph.
long[]getTimestampsPerPixel()
Calculates timestamps which correspond to individual pixels on the graph based on the graph width set with a DataProcessor method call.
double[]getValues(String sourceName)
Returns calculated values for a single datasource.
double[][]getValues()
Returns an array of all datasource values for all datasources.
double[]getValuesPerPixel(String sourceName, int pixelCount)
Method used to calculate datasource values which should be presented on the graph based on the desired graph width.
double[]getValuesPerPixel(String sourceName)
Method used to calculate datasource values which should be presented on the graph based on the graph width set with a DataProcessor method call.
booleanisPoolUsed()
Returns boolean value representing RrdDbPool usage policy.
static voidmain(String[] args)
Cute little demo.
voidprocessData()
Method that should be called once all datasources are defined.
voidsetFetchRequestResolution(long fetchRequestResolution)
Sets desired RRD archive step in seconds to be used internally while fetching data from RRD files.
voidsetPixelCount(int pixelCount)
Sets the number of pixels (target graph width).
voidsetPoolUsed(boolean poolUsed)
Sets the RrdDbPool usage policy.
voidsetStep(long step)
Roughly corresponds to the --step option in RRDTool's graph/xport commands.

Field Detail

DEFAULT_PIXEL_COUNT

public static final int DEFAULT_PIXEL_COUNT
Constant representing the default number of pixels on a JRobin graph (will be used if no other value is specified with setStep() method.

DEFAULT_POOL_USAGE_POLICY

public static final boolean DEFAULT_POOL_USAGE_POLICY
Constant that defines the default RrdDbPool usage policy. Defaults to false (i.e. the pool will not be used to fetch data from RRD files)

Constructor Detail

DataProcessor

public DataProcessor(long t1, long t2)
Creates new DataProcessor object for the given time span. Ending timestamp may be set to zero. In that case, the class will try to find the optimal ending timestamp based on the last update time of RRD files processed with the processData method.

Parameters: t1 Starting timestamp in seconds without milliseconds t2 Ending timestamp in seconds without milliseconds

Throws: RrdException Thrown if invalid timestamps are supplied

DataProcessor

public DataProcessor(Date d1, Date d2)
Creates new DataProcessor object for the given time span. Ending date may be set to null. In that case, the class will try to find optimal ending date based on the last update time of RRD files processed with the processData method.

Parameters: d1 Starting date d2 Ending date

Throws: RrdException Thrown if invalid timestamps are supplied

DataProcessor

public DataProcessor(Calendar gc1, Calendar gc2)
Creates new DataProcessor object for the given time span. Ending date may be set to null. In that case, the class will try to find optimal ending date based on the last update time of RRD files processed with the processData method.

Parameters: gc1 Starting Calendar date gc2 Ending Calendar date

Throws: RrdException Thrown if invalid timestamps are supplied

Method Detail

addDatasource

public void addDatasource(String name, Plottable plottable)

Adds a custom, plottable datasource (PDEF). The datapoints should be made available by a class extending Plottable class.

Parameters: name source name. plottable class that extends Plottable class and is suited for graphing.

addDatasource

public void addDatasource(String name, String rpnExpression)

Adds complex source (CDEF). Complex sources are evaluated using the supplied RPN expression.

Complex source name can be used:

JRobin supports the following RPN functions, operators and constants: +, -, *, /, %, SIN, COS, LOG, EXP, FLOOR, CEIL, ROUND, POW, ABS, SQRT, RANDOM, LT, LE, GT, GE, EQ, IF, MIN, MAX, LIMIT, DUP, EXC, POP, UN, UNKN, NOW, TIME, PI, E, AND, OR, XOR, PREV, PREV(sourceName), INF, NEGINF, STEP, YEAR, MONTH, DATE, HOUR, MINUTE, SECOND, WEEK, SIGN and RND.

JRobin does not force you to specify at least one simple source name as RRDTool.

For more details on RPN see RRDTool's rrdgraph man page.

Parameters: name source name. rpnExpression RPN expression containig comma (or space) delimited simple and complex source names, RPN constants, functions and operators.

addDatasource

public void addDatasource(String name, String defName, String consolFun)

Adds static source (SDEF). Static sources are the result of a consolidation function applied to *any* other source that has been defined previously.

Parameters: name source name. defName Name of the datasource to calculate the value from. consolFun Consolidation function to use for value calculation

addDatasource

public void addDatasource(String name, String file, String dsName, String consolFunc)

Adds simple datasource (DEF). Simple source name can be used:

Parameters: name source name. file Path to RRD file. dsName Datasource name defined in the RRD file. consolFunc Consolidation function that will be used to extract data from the RRD file ("AVERAGE", "MIN", "MAX" or "LAST" - these string constants are conveniently defined in the ConsolFuns class).

addDatasource

public void addDatasource(String name, String file, String dsName, String consolFunc, String backend)

Adds simple source (DEF). Source name can be used:

Parameters: name Source name. file Path to RRD file. dsName Data source name defined in the RRD file. consolFunc Consolidation function that will be used to extract data from the RRD file ("AVERAGE", "MIN", "MAX" or "LAST" - these string constants are conveniently defined in the ConsolFuns class). backend Name of the RrdBackendFactory that should be used for this RrdDb.

addDatasource

public void addDatasource(String name, FetchData fetchData)
Adds DEF datasource with datasource values already available in the FetchData object. This method is used internally by JRobin and probably has no purpose outside of it.

Parameters: name Source name. fetchData Fetched data containing values for the given source name.

dump

public String dump()
Dumps timestamps and values of all datasources in a tabelar form. Very useful for debugging.

Returns: Dumped object content.

Throws: RrdException Thrown if nothing is calculated so far (the method processData was not called).

get95Percentile

public double get95Percentile(String sourceName)
This method is just an alias for getPercentile method.

Used by ISPs which charge for bandwidth utilization on a "95th percentile" basis.

The 95th percentile is the highest source value left when the top 5% of a numerically sorted set of source data is discarded. It is used as a measure of the peak value used when one discounts a fair amount for transitory spikes. This makes it markedly different from the average.

Read more about this topic at Rednet or Bytemark.

Parameters: sourceName Datasource name

Returns: 95th percentile of fetched source values

Throws: RrdException Thrown if invalid source name is supplied

getAggregate

public double getAggregate(String sourceName, String consolFun)
Returns single aggregated value for a single datasource.

Parameters: sourceName Datasource name consolFun Consolidation function to be applied to fetched datasource values. Valid consolidation functions are MIN, MAX, LAST, FIRST, AVERAGE and TOTAL (these string constants are conveniently defined in the ConsolFuns class)

Returns: MIN, MAX, LAST, FIRST, AVERAGE or TOTAL value calculated from the data for the given datasource name

Throws: RrdException Thrown if invalid datasource name is specified, or if datasource values are not yet calculated (method processData was not called)

getAggregates

public Aggregates getAggregates(String sourceName)
Returns all (MIN, MAX, LAST, FIRST, AVERAGE and TOTAL) aggregated values for a single datasource.

Parameters: sourceName Datasource name

Returns: Object containing all aggregated values

Throws: RrdException Thrown if invalid datasource name is specified, or if datasource values are not yet calculated (method processData was not called)

getEndingTimestamp

public long getEndingTimestamp()
Returns ending timestamp. Basically, this value is equal to the ending timestamp specified in the constructor. However, if the ending timestamps was zero, it will be replaced with the real timestamp when the processData method returns. The real value will be calculated from the last update times of processed RRD files.

Returns: Ending timestamp in seconds

getFetchRequestResolution

public long getFetchRequestResolution()
Returns desired RRD archive step (reslution) in seconds to be used while fetching data from RRD files. In other words, this value will used as the last parameter of RrdDb.createFetchRequest() method when this method is called internally by this DataProcessor.

Returns: Desired archive step (fetch resolution) in seconds.

getLastRrdArchiveUpdateTime

public long getLastRrdArchiveUpdateTime()
Returns time when last RRD archive was updated (all RRD files are considered).

Returns: Last archive update time for all RRD files in this DataProcessor

getPercentile

public double getPercentile(String sourceName)
Used by ISPs which charge for bandwidth utilization on a "95th percentile" basis.

The 95th percentile is the highest source value left when the top 5% of a numerically sorted set of source data is discarded. It is used as a measure of the peak value used when one discounts a fair amount for transitory spikes. This makes it markedly different from the average.

Read more about this topic at Rednet or Bytemark.

Parameters: sourceName Datasource name

Returns: 95th percentile of fetched source values

Throws: RrdException Thrown if invalid source name is supplied

getPercentile

public double getPercentile(String sourceName, double percentile)
The same as getPercentile but with a possibility to define custom percentile boundary (different from 95).

Parameters: sourceName Datasource name. percentile Boundary percentile. Value of 95 (%) is suitable in most cases, but you are free to provide your own percentile boundary between zero and 100.

Returns: Requested percentile of fetched source values

Throws: RrdException Thrown if invalid sourcename is supplied, or if the percentile value makes no sense.

getPixelCount

public int getPixelCount()
Returns the number of pixels (target graph width). See DataProcessor for more information.

Returns: Target graph width

getSourceNames

public String[] getSourceNames()
Returns array of datasource names defined in this DataProcessor.

Returns: array of datasource names

getStep

public long getStep()
Returns the time step used for data processing. Initially, this method returns zero. Once processData is finished, the method will return the real value used for all internal computations. Roughly corresponds to the --step option in RRDTool's graph/xport commands.

Returns: Step used for data processing.

getTimestamps

public long[] getTimestamps()
Returns consolidated timestamps created with the processData method.

Returns: array of timestamps in seconds

Throws: RrdException thrown if timestamps are not calculated yet

getTimestampsPerPixel

public long[] getTimestampsPerPixel(int pixelCount)
Calculates timestamps which correspond to individual pixels on the graph.

Parameters: pixelCount Graph width

Returns: Array of timestamps

getTimestampsPerPixel

public long[] getTimestampsPerPixel()
Calculates timestamps which correspond to individual pixels on the graph based on the graph width set with a DataProcessor method call.

Returns: Array of timestamps

getValues

public double[] getValues(String sourceName)
Returns calculated values for a single datasource. Corresponding timestamps can be obtained from the getTimestamps method.

Parameters: sourceName Datasource name

Returns: an array of datasource values

Throws: RrdException Thrown if invalid datasource name is specified, or if datasource values are not yet calculated (method processData was not called)

getValues

public double[][] getValues()
Returns an array of all datasource values for all datasources. Each row in this two-dimensional array represents an array of calculated values for a single datasource. The order of rows is the same as the order in which datasources were added to this DataProcessor object.

Returns: All datasource values for all datasources. The first index is the index of the datasource, the second index is the index of the datasource value. The number of datasource values is equal to the number of timestamps returned with getTimestamps method.

Throws: RrdException Thrown if invalid datasource name is specified, or if datasource values are not yet calculated (method processData was not called)

getValuesPerPixel

public double[] getValuesPerPixel(String sourceName, int pixelCount)
Method used to calculate datasource values which should be presented on the graph based on the desired graph width. Each value returned represents a single pixel on the graph. Corresponding timestamp can be found in the array returned from getTimestampsPerPixel method.

Parameters: sourceName Datasource name pixelCount Graph width

Returns: Per-pixel datasource values

Throws: RrdException Thrown if datasource values are not yet calculated (method processData was not called)

getValuesPerPixel

public double[] getValuesPerPixel(String sourceName)
Method used to calculate datasource values which should be presented on the graph based on the graph width set with a DataProcessor method call. Each value returned represents a single pixel on the graph. Corresponding timestamp can be found in the array returned from getTimestampsPerPixel method.

Parameters: sourceName Datasource name

Returns: Per-pixel datasource values

Throws: RrdException Thrown if datasource values are not yet calculated (method processData was not called)

isPoolUsed

public boolean isPoolUsed()
Returns boolean value representing RrdDbPool usage policy.

Returns: true, if the pool will be used internally to fetch data from RRD files, false otherwise.

main

public static void main(String[] args)
Cute little demo. Uses demo.rrd file previously created by basic JRobin demo.

Parameters: args Not used

Throws: IOException RrdException

processData

public void processData()
Method that should be called once all datasources are defined. Data will be fetched from RRD files, RPN expressions will be calculated, etc.

Throws: IOException Thrown in case of I/O error (while fetching data from RRD files) RrdException Thrown in case of JRobin specific error

setFetchRequestResolution

public void setFetchRequestResolution(long fetchRequestResolution)
Sets desired RRD archive step in seconds to be used internally while fetching data from RRD files. In other words, this value will used as the last parameter of RrdDb.createFetchRequest() method when this method is called internally by this DataProcessor. If this method is never called, fetch request resolution defaults to 1 (smallest possible archive step will be chosen automatically).

Parameters: fetchRequestResolution Desired archive step (fetch resoltuion) in seconds.

setPixelCount

public void setPixelCount(int pixelCount)
Sets the number of pixels (target graph width). This number is used only to calculate pixel coordinates for JRobin graphs (methods getValuesPerPixel and getTimestampsPerPixel), but has influence neither on datasource values calculated with the processData method nor on aggregated values returned from getAggregates and similar methods. In other words, aggregated values will not change once you decide to change the dimension of your graph.

The default number of pixels is defined by constant DEFAULT_PIXEL_COUNT and can be changed with a DataProcessor method.

Parameters: pixelCount The number of pixels. If you process RRD data in order to display it on the graph, this should be the width of your graph.

setPoolUsed

public void setPoolUsed(boolean poolUsed)
Sets the RrdDbPool usage policy.

Parameters: poolUsed true, if the pool should be used to fetch data from RRD files, false otherwise.

setStep

public void setStep(long step)
Roughly corresponds to the --step option in RRDTool's graph/xport commands. Here is an explanation borrowed from RRDTool:

"By default rrdgraph calculates the width of one pixel in the time domain and tries to get data at that resolution from the RRD. With this switch you can override this behavior. If you want rrdgraph to get data at 1 hour resolution from the RRD, then you can set the step to 3600 seconds. Note, that a step smaller than 1 pixel will be silently ignored."

I think this option is not that useful, but it's here just for compatibility.

Parameters: step Time step at which data should be fetched from RRD files. If this method is not used, the step will be equal to the smallest RRD step of all processed RRD files. If no RRD file is processed, the step will be roughly equal to the with of one graph pixel (in seconds).