PhixFlow Help

Skip to end of metadata
Go to start of metadata

You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 59 Next »

A pipe is a connector that links two elements in a PhixFlow model and sends data from the input to the output. Pipes allows you to control which attributes and which records from the input are delivered by to the output, although in most cases - with minimal configuration - you will get all columns and the records from the current run.

The pipe must be enabled to make it active.

For advanced configuration, see Advanced Pipe Configuration.

A pipe joining a datasource to a data collector has no details to edit. All the configuration for the output data set occurs in the collector - either a database collector for a database datasource, or an HTTP collector for an HTTP datasource.

Name

The name is used to refer to the pipe in other model elements.

The name can have no special characters except the underscore character '_', it has to start with a letter and cannot be an Attribute Function name.

Type

There are 3 options available:

  • Pull: pull pipes are the most common type in PhixFlow - they "pull" data from the input to the output. Pull pipes are shown as solid arrows on models.

  • Look-up: look-up pipes are used to enrich data. Typically, you will have one of more pull pipes to supply the base data for an output, and if needed one or more look-up pipes to enrich the base data with values from additional inputs. Look-up pipes are shown as dashed lines on models.

  • Push: data is "pushed" rather than "pulled" into the output stream. Push pipes are most commonly used when sending data from streams to exporters (File Exporters, Database ExportersHTTP Exporters). Push pipes are shown as dotted lines on models.

Data To Read

Specify what input data to use. There are 4 options available:

  • Latest: supply data from the current run (the latest stream set). This is the mostly commonly used option.

  • Previous: supply data from the previous run (the previous stream set). This is used when you are comparing data for the current run with data from the previous run, for example, today's data with yesterday's.

  • All: supply data from all runs (all stream sets). 

  • Custom: If you are setting up a transactional model, choose this option so that you can select Only collector from same run. There are other fields revealed by selecting this option (see Advanced Pipe Configuration), but you are recommended not to update these unless directed to by PhixFlow consultants or support.


If the Only collect from same run flag is ticked, the pipe will only collect data from inputs in the same analysis run. This is only used when building a transactional model, and this configuration support several analysis runs going on at the same time without interfering with each other.

Advanced considerations - future stream sets

In some circumstances the input Stream may have Stream Sets that have dates in the future relative to the Stream Set being generated for the output Stream. This may happen, for example, if you have rolled back a number of Stream Sets on the output Stream but have not rolled back the corresponding Stream Sets on the input Stream, and have then requested that the output Stream is brought up to date. Some of the Stream Sets on the input Stream will have dates in the future relative to some of the Stream Sets you are rebuilding.

By default, pipes will ignore any Stream Sets with dates in the future relative to the Stream Set you are generating. This is so that if you are rebuilding an old Stream Set the pipe will retrieve the same data on the rerun as it retrieved when the Stream Set was first built.

Similarly, if you are running a Transactional Stream, it is possible that while your analysis run is taking place, other analysis runs which started after yours may have completed before yours. These will have generated additional Stream Sets on the input Stream with a future data relative to the date of the Stream Set you are generating. For Transactional input Streams it is possible to tell the pipe not to ignore these future Stream Sets by ticking the Read Future Data tick box on the Advanced tab.

Static

Normally when a pipe requests data from a non-static input stream, that stream will first attempt to bring itself up to date, generating new stream sets as necessary, before supplying the data requested. However, if this field is ticked, the input stream will not run.

Mandatory

If ticked, when multiple Streams are being merged then there must be an input record from this Pipe for an output record to be generated by the output Stream.

Advanced considerations - force multiple analysis runs

If this is a push pipe with positive offsets and this flag is ticked then the notification to create another stream set will only be pushed along the pipe if the last stream set created contains at least one record.

Multiplier

This causes the pipe to present each candidate set to the output stream in a different way than usual. The multiplier flag is on the Advanced tab of the form.

For each output record generated by a stream, the stream will get a set of records from each of its input pipes. If the multiplier flag is ticked on one of these, then the stream will generate an output record for each record from the set of records provided by the multiplier pipe. For each output record, each of the other input pipes will provide the same set of records as normal.

Filters, sorting and grouping, aggregating

Filters, sorting and grouping, and aggregating are configured through their own sections on the form:

FieldDescription
FilterAllows the user to set up a filter on the pipe. Also allows to set the flag to Include Audit Records. If not set, superseded records will be filtered out.
Sort/GroupSpecify the group/ order by attributes on the pipe.
Aggregate attributesSpecify any aggregate attributes on the pipe.
AdvancedConfigure advanced features on the pipe.

Filter Editor

The data being delivered by a pipe can be filtered.

Filters are made up of a set of clauses; each clause in turn contains a number of conditions. These conditions must be satisfied for data to be passed through the pipe.

Form Icons

The form provides the following buttons:

Add a clause or condition.

Delete a clause or condition.

Specifies that the value entered is a literal value. Click this icon to change this - to specify that the value entered be evaluated as an expression.

Specifies that the value entered is a evaluated as an PhixFlow expression. Click this icon to change this - to specify that the value entered be treated as a literal. Note : ["123", "234", "345"] looks like a literal value but it can be evaluated as an expression.

Open the expression in a larger editor.

Filter on Current User

Sometimes when running analysis you want to select, from the source, only records belonging to the currently logged in user. To set a filter where, say, an attribute in the source Owner equals the current logged in user, add a condition to the filter like this:

Owner Equals _user.name fx

Enter a list of values for an "Is In" or "Is Not In" filter

If you want to based on a list of values, use the Is in or Is not in comparators, then type the list of values into the comparison field as a comma separated list like this:

Country Is in England, France, Germany ABC

In this case you must NOT click the ABC icon to convert the value to an fx, because this will indicate that the value is a formula; it must be left as a literal value. If you do click the ABC icon, then the value must be entered like this:

Country Is in ["England","France","Germany"] fx

Cache Extraction Filter

A cache extraction filter allows you to further filter the data retrieved by a pipe. These are not commonly used, but are sometimes helpful when either:

  1. Optimising performance on a lookup pipe when for a set of records, the record you require from the lookup depends on non-key data, e.g. the date
  2. When getting data from a pull pipe when the filter requires that you compare one value in each record with another; this is not possible within a standard filter.

For case 1, when using a lookup pipe, data retrieved is stored in a cache. See cache size for details. The cache extraction filter allows you, as you are processing a set of output records, to use different cached entries from the lookup for each of the records are you are processing. This is very fast compared to looking up from the source (i.e. going back to an external DB table or even another PhixFlow stream) for each output record.

E.g. you want to look up the credit rating for a customer for a set of transactions - in the output, each transaction is represented by a single output record.  You create an indexed lookup pipe using CustNo as the key for the index. This means that for each new CustNo you encounter in the data, all the credit rating entries for that CustNo would be retrieved by the pipe and placed into the cache. The credit rating for each customer is fully historied, so you get a number of entries for each CustNo. To get the relevant lookup entry for each output report (each transaction), you need to compare the transaction date of the output record to the dates of credit rating entries in the cache. So to extract the relevant record, you include a cache extraction filter in the form:

StartDate >= _out.TransDate && (EndDate <= _out.TransDate || EndDate == _NULL)

Cache extraction filters are entered free hand.

The attribute names referenced must exist in a stream. This means that the each attribute must be one of:

  1. an attribute in a source stream, if you are reading from a stream
  2. if you are reading from an external database table, one of the fields returned by the database collector AND an attribute in the output stream - i.e. to use an attribute with the source as a database collector, there must be an attribute of matching name in the output stream
  3. an attribute in the destination stream, in which case you will refer to it using the format _out.AttributeName

Group/Order By Attribute

A Pipe can be grouped and sorted by attributes of the input stream. These are set up in the Group/Order section of the Pipe form. In fact, this section is called Sort/Group for pull and push pipes, and Order/Index for lookup pipes.

The following fields are configured at the level of the pipe:

FieldDescription
Maximum Number of Records Per GroupIf a value is entered into this field, then when PhixFlow collates the input records into groups according to the specified Grouping Attribute Details (see below) once the maximum number of records have been added to the group any additional records for this group will be discarded. Records will be added to the group according to the specified sort order. This can be useful, for example, where you know that for a given set of grouping attributes you may get multiple records but you are only interested in the most recent record for that set. In this case you can configure the pipe to group by the grouping attributes, sort by an appropriate date attribute in descending order and set the Maximum Number of Records to 1. This will ensure that you only get the most recent record for the specified group when you read from this pipe.
Index Type

This setting is only available for lookup pipes.

Look-up pipes can be configured for fast "indexed" access to cached data collected from external tables, files or from other streams. Indexed access is controlled through configuring a pipe with an index and setting index expressions on grouping attributes. If the Type field on the Pipe is set to 'Look-up' then the field "Index Type" becomes available. This can have the value "None" meaning that there are no index keys or "Exact Match", "Best Match" or "Near Match" as described below:

  • Exact Match: The pipe will retrieve data from its cache based on an exact match look-up with the values provided after evaluating the index expressions on the "Group By" attributes.
  • Best Match: The pipe will retrieve data from its cache based on a "Best Match" look-up after evaluating the index expressions on the "Group By" attributes. Note that the last Group By Attribute with a key expression is used for the best match lookup. The index keys on any Group By attributes with a lower sequence number are used as an initial "Exact Match" to find the set of data on which to do the "Best Match". The "Best Match" is defined as the longest key value which matches the evaluated index expression.
  • Near Match: The pipe will retrieve data from its cache based on a "Near Match" look-up after evaluating the index expressions on the "Group By" attributes. Note that the last Group By Attribute with a key expression is used for the near match lookup. The index keys on any Group By attributes with a lower sequence number are used as an initial "Exact Match" to find the set of data on which to do the "Best Match". When "Near Match" is selected, an additional field appears where you can enter an expression which should evaluate to a number representing the allowed number of edits (e.g. deletions, insertions, substitutions and transpositions) which can be made when comparing the result of the index expression to the index key in order to achieve a match.
    For example if the index key is "Smyhte" and the result of the index expression is "Smith" this would still be a match providing that the allowed number of edits is 3 or more (i.e. substitute the 'i' for a 'y', transpose the 't' and the 'h' and then insert an 'e' at the end).

Form: Grouping Attribute Details

The following fields are configured for each grouping attribute:

FieldDescription
AttributeName of an attribute in the input stream.
OrderThe position of this attribute in the list of sorting/grouping attributes.
DirectionThe direction of the sort based on this attribute: Ascending; or Descending.
GroupIf this attribute is part of the candidate key set, the Group flag must be ticked. Otherwise, the attributes will be used only to sort the data in the candidate set.
Index Expression

This field is only available for lookup pipes where you have selected indexing.

If the pipe is configured as a Look-up with an index match type set, this field becomes available. Look-up pipes can be configured for fast "indexed" access to cached data. This data is collected from external tables, files or from other streams. Indexed access is controlled through configuring a pipe with an index and setting index expressions on "Group By" attributes here.

Aggregate Attributes

Aggregate Attributes define the aggregated properties that are available when data is read from an aggregating Pipe. Note that Aggregate Attributes are not available on Pipes from Database Collectors (any aggregation can be performed in the query SQL), nor are they available on Pipes from File Collectors.

Possible aggregate values are counts, summations, averages and maximum or minimum values of Stream Items grouped in the Group/Order tab of the Pipe.

FieldDescription
Stream FunctionThe Aggregate Function e.g. Count or Sum.
AttributeThe name of the Stream Attribute to be aggregated. Note that the value in this field is not used if the Aggregate Function is Count.
NameA new name for the aggregated attribute. Note that this can be the same as the original Attribute.
OrderThe order of the aggregate attribute.

Only aggregate attributes which can be aggregated. For example, do not try to sum an attribute which contains text.

Advanced Pipe Configuration

Data to Read = Custom

The following fields are available in the Basic Settings section if you set Data To Read = Custom:

FieldDescription
From Date OffsetThe offset applied to the start of the collection period, relative to the period in the output stream that requires populating.
To Date OffsetThe offset applied to the end of the collection period, relative to the period in the output stream that requires populating.
Max Stream SetsIn almost all cases this specifies the number of stream sets to be retrieved from the input stream. However, if this is a push pipe with positive offsets this value indicates the maximum number of stream sets that can be created i.e. the maximum number of cycles this pipe can initiate.
Only collect from same runEvery time the analysis engine runs, all of the stream sets that are created by all of the streams affected by that analysis run are given the same Run ID. If this flag is ticked then the pipe will only collect stream sets from the input stream that have the same Run ID as the stream set currently being created by the output stream. You should only use this flag is both the input and output streams are transactional.
HistoriedIf ticked, the pipe will collect data from the input stream by period. So if the from and to date offsets are both 0.0, and the output stream requires stream generation for the period 17/10/07 - 18/10/07, data will be collected from the input stream for the period 17/10/07 - 18/10/07. If not ticked, all data will be collected from the input stream, regardless of period. In this case, the offsets are still used to determine whether the required data periods in the input stream exist before the stream calculation can be carried out.
Read Future Data

If you are running a transactional stream then it is possible that while your analysis run is taking place, other analysis runs which started after yours may have managed to complete before yours, generating additional stream sets on the input stream. These additional stream sets will then have a future data relative to the date of the stream set you are generating. By default PhixFlow will ignore input stream sets that have a date in the future relative to the stream set being generated.

However, for transactional streams it is possible to tell the pipe to include future stream sets by ticking this box.

This field is only available if the input stream is transactional and:

  • Only collect from same run is not ticked
  • Max Stream Sets is blank or zero
  • Historied is not ticked

Advanced section

The following fields are configured in the Advanced section:

FieldDescription
Data ExpectedThis field is available when the Pipe Type is Push or Pull. This flag allows the user to specify that the pipe is expecting to receive data. If ticked but no data is received this is treated as an error.
Allow Incomplete Stream Sets

Normally, when a pipe tries to read from an input stream that contains an incomplete stream set, PhixFlow will attempt to complete the stream set before passing data down the pipe. However if the stream is static (i.e. the stream has its 'static' flag ticked) or is effectively static (i.e. all of the pipes reading from it in this analysis run are static) then, instead of completing the stream set, an error message is produced indicating that you cannot read from this stream because it contains an incomplete stream set.


If you do not want this error message to be produced when reading from static (or effectively static) streams, but would instead prefer PhixFlow to ignore the incomplete stream sets, then you must tick this box on all pipes that will read from the input stream in this analysis run. If there are multiple pipes that read from the input stream during this analysis run and even one of the pipes does not have this box ticked then you will not be allowed to read from the stream and the error message will be produced.


Pipes which are not used in the current analysis run (for example where they lead to streams on branches of the model which are not run by the current task plan) have no effect on whether or not the error message is produced.

Cache Size

The cache is used when carrying out lookups from streams or database collectors. When doing a lookup, there are two common scenarios:

  1. The pipe does a single lookup onto a stream or database table to get a large number of records in one go (e.g. 10,000 records)
  2. The pipe does many lookups, getting a small number of records for each lookup (e.g. 10 records at a time).

In case 2, the results returned are typically based on a key value, e.g. an account number. This will be used in the filter of the pipe, if you are reading from a stream, or in the query, if you are reading from a database collector. For example, the query in a database collector will include the condition:

WHERE AccountNumber = _out.AccountNum

For efficiency, the records are cached (stored temporarily in memory) so that if the same set of records need to be looked up again they are readily available without going back to the database.

This field allows you to set a limit on the size of the cache. Setting a limit is important because if you do not, the cache can become very large and consume a lot of memory, which can lead to a slow down in both your tasks and those of other users of PhixFlow.

To set the cache size, try to estimate the largest number of records that the lookup pipe will return on a single read.

If you do not set a limit, it will default to the system-wide default, specified in the Maximum Pipe Cache Size in the System Tuning tab of the System Configuration.

Cache warnings and errors

If a single read brings back over 90% of the specified cache size, a warning message will be logged to the console.

If a single read brings back 100% or more of the cache size, a second warning message will be generated. If the Enforce Cache Size limit flag is ticked in System Configuration, instead of a warning an error will be generated, and the analysis run will stop completely.

Error Message: Cache Size Limit Exceeded
The Pipe "stream_name.lookup_pipe_name" cache is 100% full (the cache size is 10).
Technical breakout
Every time the lookup pipe is referenced, PhixFlow calculates the values of all of the variable elements of the query or pipe filter, and checks if it already has a set of data in the cache retrieved using this set of variable values. If so the data is immediately returned from the cache. Otherwise, a new set of data is read from the stream of collector. If adding the new records to the cache would cause it to exceed the maximum cache size, previously cached results are removed to make enough room for the new results.
Buffer SizeThe buffer size used to perform the stream calculation. If a large amount of data is being processed, then setting a large buffer size will give better performance.
Pipe ViewThe pipe view is used to limit which fields are retrieved down the pipe and in what order, and in some circumstances how each field is to be formatted. You can select from any of the views that have been configured on the source stream. Please note that any sorting or filtering of records will have to be applied directly on the pipe, and will not be inherited from the pipe view.

The pipe view is used in two contexts.

During lookups

Pipe views can also be used on lookup pipes to limit the fields that are returned by the lookup request. This is most useful in the scenario where the you want to read and cache data on a lookup pipe from a stream that has lots of attributes but where only a small number of attributes are actually required. You can simply create a new view on the source stream listing only the attributes needed, then specify it as the pipe view on the lookup pipe. Only those attributes specified on the view will then be loaded.

Without a pipe view, the pipe will load and cache all of the attributes from the stream which may consume a significant amount of free memory if there are a large number of records.

During File Export

When sending data to a file exporter only those fields specified on the pipe view will be exported. If no pipe view is supplied then all fields will be exported.

If the file exporter is configured to export to Excel or to HTML, and no Excel template is specified on the exporter, then the pipe view will be checked to see if an Excel template has been specified there. This template will then be used as part of the export. See the description of file exporters for further details. If an Excel template has been specified on the exporter then this will override any template specified on the pipe view.

If the file exporter is configured to export to HTML and the pipe view is a chart view then the output will be a picture of the chart in PNG format.

Max Records To Read


The maximum number of records that should be read down this pipe. The pipe may read more than this number of records if it is configured to carry out multiple reads simultaneously e.g. if it is connected to a File Collector which has been configured to read multiple files simultaneously or if this pipe strategy is "Directed" with multiple workers.
Strategy

The Execution Strategy determines how this pipe should be implemented. See the section on Directed Merge Strategy

Max Workers

This field is only available if Strategy = Directed

The maximum number of concurrent worker tasks.

If blank, this defaults to 1.

Worker Size

This field is only available if Strategy = Directed

The number of key values to read for a single worker task (which runs a single select statement).

If blank, this defaults to 1000. This is the maximum value that can be used when reading from an Oracle database.

  • No labels