Runtime Engine Server Resource

Transactions and transaction groups have a hierarchical tree structure in which a desired transaction can be found by parsing a request message. A transaction group is represented as a parent directory that contains sub transaction groups and transactions. Business resources related to the transaction group are in the .iar archive file, just like a transaction. Transaction and transaction group information are in the XML document with the .biztx extension in the .iar archive file.

A transaction consists of a request message, a response message, and an abnormal response message, and inherits the information of transaction groups contained in the parent node.

A transaction group that is located in the parent node or the intermediate node of a transaction tree can have a sub-transaction identifier to find a sub transaction group or a transaction node (which is the end node) by parsing a request message.

The basic purpose of a transaction tree is to find a transaction that matches the information of a message by parsing a request message and to invoke a service (service flows, outbound rules, or multi-bound rules) of the AnyLink engine.

Transaction and transaction group information contain symbolic link information that is referenced by a transaction node or a transaction group node. A symbolic link links other transaction nodes that contain the resources referenced by the transaction node. Since basically only resources in a parent node can be referenced, a symbolic link is useful when sharing resources in transaction or transaction group nodes other than the parent node.

Among the resources that are managed together in AnyLink transactions or transaction group nodes, there are various resources that are converted to Java code. For example, messages, mappings, expressions, and various user classes are resources that can be used in Java code.

Java code is read through a class loader, and each AnyLink transaction tree is designed to have parent and child relationships in the class loader.

A symbolic link node is not a parent on a class loader, but the node can be used as if it were a parent thanks to the AnyLink ClassLoader implementation.

A service flow definition file is drawn by using a BPMN (Business Process Modelling Notation) diagram in AnyLink Studio’s service flow editor.

Service flows are managed in the service flow engine, and the message events that are received are registered as AnyLink’s internal services.

In the most basic form of AnyLink transactions, messages are typically received through an inbound adapter, and the messages are parsed through a transaction tree for identifying internal services. Typically, an internal service of AnyLink is a message event of a service flow, and it proceeds according to the flow defined in the service flow, and then calls the service activity of the service flow. Service activities drawn in a flow are mostly adapter outbound rule services.

The following describes the most basic form of an AnyLink transaction: an external system is called though an outbound rule service and a response is received. The response is then passed to the calling flow and the response message is sent to the adapter that sent the initial message.

Parsing rules, which are rules for parsing external messages, are not managed as separate resources but instead are defined in transactions and transaction groups.

A document that defines outbound adapter rules used to send requests from AnyLink to other systems. The configuration information differs according to the protocol and implementation of each adapter. Basic configuration information consists of request messages, response messages, abnormal response messages, and adapter and endpoint IDs.

An outbound rule is one of AnyLink’s internal services. If an outbound rule service is called internally from another component, it identifies the adapter and endpoint that are set in the rule, and sends messages to an external system.

If an internal service of AnyLink needs to be routed differently depending on the situation, multiple services can be grouped together as a single group. This type of service is a multi-binding router rule. The following can be configured: service splitting, round robin, weight-based or time-based round robin, service flow correlation, handlers, etc. Request, response, and abnormal response message mapping (message conversion) can also be specified.

Multiple messages can exist inside a .Umsg file. AnyLink distinguishes two different types of messages: DTO (Data Transfer Object) which expresses the internal structure of the data used for input and output, and messages in specific formats. A .Umsg file can contain a single DTO and messages in many different formats. The messages must have the same structure with the DTO. The supported message formats include XML, fixed length, delimiter, NameValue, and JSON.

AnyLink converts DTOs and messages into Java code and manages it. This is for optimizing performance and efficiency. This means that the .class files generated by compiling .Java source files that correspond to .Umsg documents are included in transaction resources.

A message mapping document defines the rules for converting input messages to output messages. AnyLink specifies conversion rules in the form of connecting the fields of each input message and output message, and can use various expressions as mapping sources in addition to the input message.

To improve the execution speed of mapping that consists of a combination of input/output fields or expressions, AnyLink internally converts mapping to Java code and manages it. Hence, just like message files, the .class file generated by compiling .Java source file is included in the transaction resource.

AnyLink uses expressions to indicate a part of a message or express a function expression.

An AnyLink expression can express the structure of a message. For example, if a field named field 1 exists in msg1, it is represented by the expression "msg1.field1". In a service flow, the user can access a DataField variable which is declared in a flow. For example, to access a field named "name" of a variable named dataField1, the user uses the expression "$dataField1.name".

In an AnyLink expression, the user can use parentheses to represent groups, and call Java methods. For example, in the variable example above, if a field named "name" corresponds to the Java String class, the user can extract only the first two characters as follows.

Optionally, the user can explicitly specify the Java type of the variable. For example, if a variable named dataField1 has a datatype of com.example.NamedPerson, the user can rewrite the above expression as follows.

AnyLink converts expressions to Java code by considering the execution speed. Incorrect expressions will cause errors when converting and compiling Java code during deployment in Studio. Same as message and mapping files, .class files generated by compiling .expr files are included in transaction resources.

In AnyLink, expressions are used in splitting condition statements of service flows, correlation calculations, and multi-binding router rules.

Expressions that can be used to represent the source information of a message mapping have a slightly special form in order to support mapping characteristics, such as creating an empty object if there is an object reference of an empty source, and supporting array type mapping.

Environment configuration documents of transactions and transaction groups are not directly related to transaction information, but stores information related to the transaction environment.

Transactions and transaction group resources are managed in a single .iar file for each node, but the environment configuration document files of transaction nodes exist separately in a document format with the .bizcfg extension under the directory that indicates the transaction node or transaction group node.

The environment configuration document can set the transaction and outbound rule trace log levels. The user can also specify thread pools to be used in service flows of corresponding transaction nodes.