Essentially EAI tool is doing data integration - at all times or most of the times. Now if the integration is real time the amount/size of data these tools carry/process/transmit is small. As soon as an event occurs in a system its trasmitted to other system. For example: A warehouse inventory in legacy world would reflect the inventory which is a day old. Because they are designed in such a way. But in realistic or new world scenario you need to keep your inventory uptodate. So as soon as a sale happens in the store or on website this should reflect the correct inventory position in warehouse. I have seen cases where if a vendor fails to fulfill an order which was executed ( mainly due to wrong inventory psoition - stock over but order placed) are charged by the other party for a failed order.
In some case business model or application can change to this kind of scenario which is more real time or near real time. But in a number of cases it may not be so. Sometime or most of the times - The legacy application can not produce the data as event occurs - reason may be because legacy code/apps are interconnected and data is produced only end of day , end of month etc.
When the data volumes are high on daily / weekly/ monthly scenario and there is batch window to process them. EAI may not be suitable. It would take design and interface to process such huge data - time to process the data.
If no. of such interfaces where u only process huge files EOD, EOM and data volumes are high. ETL tool is the most efficient way and EAI tool should be avoided
Lastly: EAI & ETL co-exist in big organization : ETL -to support legacy batch data and EAI - To move to real time environment
By
jai mishra, Tech Architect tcs
| 04 08 2009 21:44:15 +0000
Mr. Jai can you elaborate on your point and explain the efficiency with which these tools handle data integration?
By
Jyoti Rath, Sr. Associate, Barclays
| 04 07 2009 09:04:41 +0000
Yes but I am wondering what was the reason behind asking this question. Are you trying to compare EAI to ETL? If you look at the premise of Integration, on the lowest level it is data integration always between two applications/systems.
By
Vipul Gupta, Information Systems(MIS)-Manager, IB
| 02 24 2009 16:18:46 +0000
yes , EAI tools like TIBCO, WebMethods etc are aimed at Data Level integration & system level integration ( A2A ) , business integration(B2B) and Business Process Management ( BPM). with work experience in TIBCO, SAP XI , webmethods I can say that EAI tools provide very good integration frameworks for data integration between two systems or between two separate business entity. TIBCO provides adapters, messaging trasports , jdbc and webservice option for doing data level integrations.
By
Ranjeet Singh, Tech Architect, TCS
| 02 13 2009 17:47:39 +0000
EAI tools can definitely do data level integration. My experience is
with Vitria, but maybe someone else can give you information on the
other EAI tools.
Vitria performs data level integration using
JDBC/ODBC drivers. You can either run SQL queries against the database
or you can retrieve data based on add/delete/update triggers. Vitria
offers off-the-shelf connectors/adaptors for most of the popular
databases (Oracle, MSSQL, Sybase, DB2, Informix).
By
Maneesh Jain, Business Analyst, Oracle
| 11 30 2008 20:40:25 +0000
|
I have worked and currently working in EAI projects in the past. When the integration is at a very low level, we had issues. The tools we used could not handle to load. But if the integration is at a relatively high level, which will result in low volume data traffic, it worked well. When the integration requirement is at a low level, I would recommend traditional integration that are built to handle volume.
By
Somasundaram Natarajana, Tech Architect, Cognizant Technology Solutions
| 03 10 2009 19:15:58 +0000
I believe the misperception of ETL being cheaper than EAI is due to ETL
being generally used for tactical solutions (narrow in scope) while EAI
solutions tend to be more strategic (broad in scope).
By
Ashish Kumar Mishra, Business Analyst, SAP
| 11 30 2008 20:41:57 +0000
|