This project contains a set of tools designed to help optimize BusinessWorks 5 projects. The current set of functionality includes:
This utility discovers BusinessWorks activities that have not been executed within BusinessWorks processes that have executed. It outputs a file of comma-separated values with two values on each line: The process name, and the activity name. Thus if a BusinessWorks process has five activities that have never been executed, there will be five lines in the result file witn the first value (process name) the same, and the second value (activity name) different.
It should be noted that not all dead code is bad. Exception-handling logic for example is absolutely necessary, however in many cases will never have been executed. Care and understanding should therefore be taken when reviewing and evaluating the results.
this class works by creating two lists for each BusinessWorks process. One list contains all activities within the process, and the other contains only those that have executed. The difference between the two sets are the un-executed processes or dead code.
This utility modifies generated value-of within if constructs for optional-to-optional mappings to straight copy-of.
We often see a construct similar to the following if we examine BW process files in a test editor:
`<xsl:if test="pfx10:ContactFrequency">
<pfx10:ContactFrequency>
<xsl:value-of select="pfx10:ContactFrequency"/>
</pfx10:ContactFrequency>
</xsl:if>`
The 'if' statement surrounding the actual mapping is often generated by Designer as the result of an optional-to-optional mapping. Where the value is being passed unchanged, the same result can be delivered much more succinctly with the following construct instead:
`<xsl:copy-of select="pfx10:ContactFrequency"/>`
This will copy the value if it exists in the source data. Otherwise it should not produce an output element. While this may seem like a "trivial" savings, where large XML documents are being mapped the actual savings can be significant.
It is important to note that the space savings that we see on disk does not necessarily translate to in-memory savings. When processing large XML documents most of the memory consumption may end up being the actual transient data being passed and manipulated rather than the mappings. So if the goal is to reduce memory utilization it may be best to consider ways that a large XML document can be broken down into smaller chunks so that the entire document does not need to be in memory at once.
This utility traverse a project directory and locate all BW process files. For each process file introspect and evaluate the imported namespaces in order to determine whether the namespaces are actually referenced.
Over the course of time as a result over normal development and maintenance activities within BW processes are changed and or removed. This often results in scenarios where schemas that were once referenced are no longer referenced.
While the additional unnecessary schema references will not cause any harm functionally, it adds additional processing and memory overhead since the schemas must be loaded even if they are not used. It is therefore a good practice to clean up unused schemas on a periodic basis. This class helps with that task.
Introspects BusinessWorks processes, and evaluates the input mappings to each activity. It is known that the cost of XML parsing increases significantly in relation to the depth of nested elements within XML instances. This problem is multiplied many-fold if there are multiple references to deeply nested XML elements within the same activity input. Each time the parser encounters another deeply nested element the cost is incurred yet again.
It is possible to significantly reduce this cost by creating shortcuts to deeply nested levels within the XML. For example, let's examine the following hypothetical example:
$ActivityOutput/ns:Accounts/ns:AccountArray/ns:BankAccounts/ns:CheckingAccounts/ns:CheckingAccount
The last element above (ns:CheckingAccount) has sub-elements called:
- ns:Name
- ns:Type
- ns:Number
By creating a shortcut called $CheckingAccount that referes to the deeply nested element we can allow the deeply nested element to be evaluated just once, rather than once for every sub-element.
This class produces a report that shows both the depth of elements that are referred to within activity input mappings, and also the number of times the deeply nested reference is itself referenced. This allows users to prioritize the effort of creating semantically meaningful shortcuts based on both depth and number of references.
Intruments a BACKUP COPY of an existing BW application with tooling that generates a heap dump after the Start activity and before theEnd activity of each process in the project.
The resulting heap dumps can be subsequently analyzed with the Eclipse Memory Analyzer Tool (MAT) from the command line, to produce reports on memory utilization and class histograms.
This allows memory growth to be analyzed within the context of BW processes such that it becomes easier to see which processes cause more memory to be consumed, and which classes are generated within the BW process that consume the memory.
After running the BusinessWorks project that was previously instrumented
using the nochum.bw5.utilities.Optimize instrument option, a directory
is populated with JVM heap histogram profile snapshots taken at the
start and end of every BW process that was invoked. The "instrument"
tool places these snapshots in the location specified by the -output
argument.
This option reads all of the JVM heap histogram profile snapshots that the "instrument tool" created, and creates a summary report that shows the total number of live objects allocated and the total memory utilization by live objects on the JVM heap at the beginning and at the end of each BusinessWorks process invocation. This allows the user to pinpoint exactly where memory spikes are taking place.
Usage: Optimize COMMAND where COMMAND is one of:
- deadcode - Discover and report dead code within a BW project.
- ifToCopyOf - Modify generated value-of within if constructs for optional-to-optional mappings to straight copy-of.
- namespaces - Remove unused namespace declarations from processes.
- xpathRef - Report on the number and depth of XPath references for each activity in each process.
- instrument - Instrument a BACKUP COPY of a BW project for memory profiling.
- histoSummary - Create a summary report of BW memory utilization by class within BW process.
Most commands print help when invoked w/o parameters
Current planned enhancements include:
- checking for HTTP, JDBC, and SOAP activities that do not have timeouts configured
- analysis of engine TRA files for sanity checks