DUPLICATE OPERATION DETECTION SUPPORT

Duplicate Operation Detection Support Overview

Some operations are idempotent in nature, i.e. they can be performed more than once without any harm. However, some other operations are non-idempotent in nature, i.e. they should be performed only once. In the case of non-idempotent operations, performer should be able to detect duplicate operations and perform each non- idempotent operation only once.

Examples of non-idempotent operations are Submission and Delivery of messages which shouldn't be performed more than once. Examples of idempotent operations are Submission-control and Delivery-control which can be performed more than once with no harm.

ESRO Services don't detect duplicate invocation of operations. As a result, the Duplicate Operation Detection Support Functional Unit is used to detect duplication when the same operation instance is invoked more than once. Invoker assigns an Operation Instance Identifier to an operation and this Operation Instance Identifier is used at the peer performer entity to detect the duplicate invocation of the same operation.

Using this support, non-idempotent operations can be repeated over and over with no harm because the duplicate invocations are detected by this functional unit. This support helps the performer not to perform an operation more than once.

Support for duplication detection is realized through allocating Operation Instance Id (see Section 4.1.2, "Operation Instance Identifier") to an operation by invoker. When an operation is invoked using duplication detection support, performer logs the Operation Instance Identifier and checks the next operations against duplication.

Operation value identifies whether performer should detect duplicate operations (see Section 4.1.1, ``Operation Value") and Operation Instance Id is assigned by invoker and sent as the first byte of operation's parameter.

Operation Value

Operation Values are divided into two groups. Operation values from 0 to 31 do not have Duplicate Operation Detection Support (0 to 31) and operation values from 32 to 63 have Duplicate Operation Detection Support.

Duplicate Operation Detection Functional Unit checks for duplication only if Operation Value is in the range of 32 to 63.

When invoker user uses an Operation Value in the range of 32 to 63 which means operation with support for duplication detection, the user should specify an Operation Instance ID for the operation (see next section).

Operation Instance Identifier

To support duplication detection, an Operation Instance Identifier is assigned by invoker user and sent as the first byte of the operation's parameter. This identifier is used on performer side to detect duplicate invocation of the same operation. Characteristics of Operation Instance Identifier is as follows:

• Operation Instance Identifier is one byte and can have values from 0 to 255.

• Operation Instance Identifier is sent as the first byte of the operations parameter (without encoding).

• The length of Operation Instance Identifier is 8-bit, but depending on the performer capabilities, it might keep 0 to 127 Operation Instance Identifiers for duplication detection. The performer profile defines the number of outstanding Operation Instance Identifiers that are checked against duplication. When a performer profile indicates support for 0 outstanding Operation Instance Identifier, it means it does not have support for Duplicate Operation Detection. In this case, there should be only one outstanding operation at any point of time.

• Instance ID check is not part of ESROS, per se. Use of Duplicate Detection is determined by EMSD-P. Operation Instance ID for operations 32-63 is the first byte of the argument. Duplicate Detection suuport strips that byte.

• The Instance ID is not subject to Basic Encoding Rules (BER).

• The invoker user assigns the Operation Instance Identifier to the operation at the time of requesting the invoke service. The Operation Value should be in the range of operation values with duplication detection support, i.e. 32 to 63.

• It's the responsibility of the user to choose Operation Instance Identifier in a way that uniqely and unambiguously identifies the operation.

• From the invoker's perspective, assumption is that two operations with the same operation Instance Identifier are totally identical which means they produce exact same results.

• Operation Instance Identifier uniqely specifies a non-idempotent operation and multiple invocations of such an operation will eventually result in the same outcome because the duplicate instances are identified and the operation is not performed more than once.

• From the performer's perspective, assumption is that two operations with the same Operation Instance Identifier should be executed once and once only.

• If requested, the degree of duplication checked by Duplicate Operation Detection Support Functional Unit on the performer's side (i.e. the total number of outstanding Operation Instance Identifier kept) can be communicated with the invoker to synchronize the invocations.

• User of Duplicate Operation Detection Support is responsible to behave based on the performer profile and its limitations in this regard. This behavior is defined based on the desired semantic of the operation which is to be implemented.

• On the performer side, when an Operation Instance Identifier is received, a previous Operation Instance Identifier whose distance to this latest one is greater than or equal to half of the wrap-around range of the Operation Instance Identifier number is expired, i.e. for an 8-bit Operation Instance Identifier, the distance of 128 causes an old Operation Instance Identifier to expire.

• It's the responsibility of the invoker user to use consecutive Operation Instance Identifier numbers, or when it skips some Operation Instance Identifiers, it should remember that if there is an smaller Operation Instance Identifier on performer side with the distance explained above, it will be expired.