How StreamServer reads and processes XML documents

The StreamServer reads an XML document starting from the top. In the collection phase, the StreamServer continuously reads, parses, and saves input data until it comes to a specific check point in the document. The check point can be an end tag, a comment, a pattern, etc. What it is depends on the current collection mode – Document, Node, or Message. When a check point is reached, the StreamServer swaps to the process phase, and processes data that is currently available in the memory. The StreamServer continues to swap between collection phase and process phase until it reaches the end tag in the XML document.

Multiple Events using the same XML document as input

The StreamServer may use different collection modes for different parts of the input XML document. It all depends on the number of Events that use the same XML document as input, and which collection modes you have specified for the Events.

Available data

When the StreamServer processes data, it has access to the data that is currently available in the memory – nothing more and nothing less.

Releasing memory

The more data occupying the memory, the less performance. The StreamServer will therefore continuously release memory when operating in Message and Node collection mode. Consequently, all removed data will be unavailable to the StreamServer in the subsequent process phases.

Collection modes

Document mode
Scope	This collection mode overrides any other collection mode, i.e. the whole XML document will be collected using the Document collection mode. The collection phase continues all the way through the XML document to the end tag for the root element. This means that the whole XML document will be available during the process phase.
Memory	No memory is released until the StreamServer has finished processing the whole XML document.
Availability vs. performance	This mode provides the best data availability since the whole XML document is available during the entire process phase. Performance will be affected because of the high level of memory usage.

 

Node mode
Scope	This is the default collection mode. The collection phase continues until the StreamServer finds an end tag, a comment, or processing instruction.
Memory	Text nodes: If an element contains more than one text node at the same level, only one of them is kept in the memory. If a second text node is found while parsing the document, the first text node is removed. For example, while parsing <A>billy<B/>bob<A/>, billy is removed when bob is parsed. Child elements: Child elements will be removed when the parent element has been processed. For example, when the StreamServer has processed the elements <A><B/><C/></A>, the child elements <B/> and <C/> will be removed.
Availability vs. performance	This mode provides the best performance because of the low level of memory usage. Data availability will be affected since only a small portion of the XML document is available during each process phase.

 

Message mode
Scope	The StreamServer runs in Node collection mode until it finds a matching pattern. When it finds a matching pattern, it swaps to Message collection mode – provided the corresponding Event is configured to use Message as collection mode. The collection phase extends from [pattern start] to [pattern end] according to the following example: <[pattern start]A><B/><C>text</C></A[pattern end]>
Memory	When the StreamServer has finished a process phase, only the main element is kept – everything else is removed. For example, after processing <A><B/><C>text</C></A>, only <A></A> is kept.
Availability vs. performance	Data availability and performance depends on the amount of data enclosed by [pattern start] and [pattern end]. For example, if the pattern is defined for the root element, the Message mode is the same as the Document mode. On the other hand, if the pattern encloses one single element, the Message mode is the same as the Node mode.

Nested Events

Multiple Events can retrieve input from the same XML document. Which parts of the document an Event will retrieve is defined by patterns. If one Event’s pattern is enclosed by another Event’s pattern, this is known as nested Events.

Example 1

Nested Events

...

<EventA>text

   <EventB>text</EventB>

</EventA>

...

If EventA operates in collection mode Message, EventB will also be forced to operate in collection mode Message. Everything enclosed by the parent pattern – in this example EventA and EventB – will be available during the process phase. This means that the StreamServer will have access to all data that belongs to EventA when it processes EventB, and vice versa.