Just Imagine…. Your ERP application recognises that your purchase order request has not been delivered to your third party vendor application due to a system shutdown. Recognizing that the PO could not be delivered, due to a third party system failure, your system creates a new Purchase Order with the same data and delivers it to the vendor application after some time and informs you that “I have created a new order because of an error in delivering the old order and here is the new order number!”
The whole thing happenes without any human intervention. This should be the ideal e-business application.Now you might ask where does applications and systems monitoring come into picture in such a scenario.
Taking the above simple example, the system shutdown raises an alert, the alert is correlated to all the processes affected during that time period, then the corrective measures for each process affected is taken automatically
Our very own Netweaver Administrator which is a part of our platform is a step in this direction. Netweaver Administrator aims to provide a uniform lifecycle management solution for all our netweaver components and in future could be used to monitor any heterogenous business landscape in general.
Looking at the enabling technologies which are currently available for such an approach purpose we have JMX from the JAVA world and WSDM from web services arena. These approaches model the resource to be managed, for example a system or an application component, into an manageable entity. This entity is in the form of a management object in the case of JMX and in the case of WSDM its a web service endpoint. Once this entity modelling is done then there are various management capabilities added around them.
Also when we talk about managing different components and applications it would be nice if all of these components communicate in a standard language or format which would enable easy understanding. For this we have standards like CIM which are still evolving.
To give you some examples of how things would work in a efficiently “managed” digital world –
- The message meant to reach an R/3 system fails because of the XI adapter being down, immediately restart of the adapter is triggered and the message is resent.
- Memory consumed by a java process is causing performance problems for the server as a whole, immediately the process is killed and restarted or a cluster is dynamically added
- In case of hardware problems, user is guided step by step to solve the issue in a user friendly approach
- Applications can be monitored more effectively – like a j2ee application can be monitored by getting the information of number of DB connections used, state of the application in the server, version, memory consumed etc
It is very interesting to think about the fact that it is only possible for IT applications and IT infrastructure to monitor and detect its own problems and also take healing measures for it on its own. A car cannot detect a problem in its engine and rectify the piston rings all on its own, but it is possible for an erp application to see that it has occupied a lot of memory and it is going to crash and automatically switch its memory space !
This is where e-business becomes eternal !