WSO2 Siddhi存储的事件/结果

Question

According to the WSO2 web site, Siddhi can handle 2.5M requests per second, which is really nice.

I am expecting to use the product for a event routing proxy. Ex: Gets requests over http/ process the event/ forward the event to know destinations.

Ex: A customer buys a 2kg of rice. POS system call an API(SOAP or REST) exposed at event processor. CEP checks whether bought weight is greater than 1kg, which is true here. Then CEP calls an API at A and an API at B.

I prefer to get to know below information about features of Siddhi for above process.

Is there a specific implementations in siddhi for below points.

1) which allows to receive SOAP/REST/ANY OTHER, calls from external systems efficiently/faster. Ex: In above example, getting the call from POS when rice is bought.

2) persist events/results of event in specific cases and process later under conditional scenario. Ex: in above example, if the API A or B is down, events or stored results should be persist and should send to the API A,B once they are up.

Thank you.

Answer 1

1. WSO2 Complex event processor has many different adaptors implementing various protocols such as HTTP/SOAP/JMS which can be used for this requirement. Refer to the documentation on input event adaptors here for more details.

2. In Siddhi, there are 'event tables' which can be used to store/retrieve/query events. They are stored in an RDBMS like MySQL. Documentation on defining an event table is here . Using an already defined event table in a query is given here . Also, Sample 0109 demonstrates a simple use case of event tables.

However, note that to implement the 2nd requirement, you may need to send some trigger event to CEP when an API is up, coz CEP needs to know when they are up.

As a side note on performance of Siddhi, it mainly depends on the type/complexity of the queries you write and you may not be able to achieve 2.5M events/sec performance for some types of queries - in this case the queries that use databases (event tables) will have a lower performance than that due to the disk access.