2017-11-15 22:34:00 UTC

Oracle ICS vs ODI


What is your recommendation in choosing an integration product for implementing fusion applications? ICS or ODI?

I have heavy duty file processing in interfacing Cloud ERP with many 3rd party systems, but not they are not voluminous. 

Which would you recommend?

Guest
1010 Answers

author avatar
Consultant

ICS and ODI are not mutually exclusive, and can be used in a tandem, ICS is good for publish and subscribe patterns, also has lots of adapters such as file, ftp and various SaaS adapters. ODI is for heavy transformations, but has almost no concept of publish and subscribe (there is pub/sub pattern within ODI but is not the same as messaging pub/sub). So, I would use ODI for heavy transformations on large files if required, ICS for integration via out of the box provided adapters.

2017-11-22 13:56:26 UTC
author avatar
Top 5Consultant

Is your business a disconnected business?
Aren't the legacy systems being synchronized?
Is your cloud systems not being synchronized with on premisses-legacy systems?
Does your business already have transactional systems such as ERP, CRM in the cloud? And you need to push or pull data from or to the cloud/premises?
Is it going to cloud a goal for your business and there is already a strategy for that? if there is then why not going with the ICS?

even though ODI can pull and push data to the cloud, depending on the cloud provider you may face issues, for example I have worked on project where the client wanted to pull data from Salesforce(cloud-based CRM) and ODI didn't have native connector, by the way, it had but it was deprecated maybe for commercial issues because Oracle also has a CRM system(Siebel) which competes with SF. To overcome that problem we had to use a custom web service, on the other hand, for example, the same client had multiple SQL server databases on Microsoft Azure which we simply treated as another server and pulled data from there to on-premise Oracle DW.

Typically the dilema many business face is the lack of a single view of the main business entities, such as "Customer", "Account" etc..which can lead to whole lot of data
discrepancies or wrong information causing an impact on the business and impacting the data-driven decision making.

ODI is a really powerful ETL/ELT tool, which can handle virtually any kind of heavy data integration, it is a Java-based application which is super flexible and can easily be customized or extended,
It handles quite well the heavy of duty of not only popular RDBMS but also flat files, XML files, JSON, web services etc... if you use SOA, for example, you can set BPEL to call ODI scenarios depending on business rules/events

On the other hand as far as I can see the "Oracle Integration Cloud" is quite powerful, being able to integrate on-premises systems with Cloud applications,
with many out-of-the-box connectors/adaptors to a wide range of technologies/platforms especially those in the cloud, but also can pull data from premises and synchronise in real time
Enabling the business to get rid of silos of data scattered across departments/systems. Data integration is vital to business to have a single view of business events that is happening in their mini-world.

To make a decision on this I would check the company's strategy/goal, bearing in mind that cloud is the future and everybody is going there, second I think there is more flexibility to integrate several data formats, structured, semi-structured and unstructured(BIG data essentially)although ODI can also handle big data with the Hive Km,
Based on my experience, the government is the type of business where there is more reluctance to go to the cloud due to sensitive data/information.

It would be better to have more information about the nature of the business and understand the main bottlenecks or areas where improvement is desired for a precise recommendation.

2017-11-23 04:31:53 UTC
author avatar
Vendor

I regret to say that I don’t use both of them.
Therefore it’s not fair to vote for one of them.

My own experience ( I developed an ETL tool for my own company dataprisma) was that many ETL – tools just
offer connectivity and interfacing - businesslogik
Most of the connectivity comes not from the ETL tool but from the middleware
(drivers). The more important part, the businesslogic concentrates very often on very easy point and click tools to connect table-entries.
But -In many cases this is not enough:
I used to work with applications (finance-systems, building best of bread architectures ) that needed a lot of sophisticated businesslogic in the etl- layer.

When Your are happy with SQL for this task, the most simple ETL – tool will be good for You.
But maybe it makes sense for your business that not Your programmers (IT) but Your business analysts (business) define and maintain
the transferlogik.
So my advice: check the amount and complexity of the transformation- logic that Your solution needs and select the easiest tool
that supports you best in formulating and maintaining your business-logik.

2017-11-23 12:18:35 UTC
author avatar
Vendor

Hi Nick,

It should be a combination of ICS and ODI. ICS is well versed for SAAS based integrations but does not work for bulk data and its maturing enough day by day whereas ODI is purely for bulk load and with small workarounds(customizations) it fits our purposes.

For your case, I would prefer ODI. If there are huge event-driven web services and integrating third-party systems, I would prefer ICS.

2017-11-23 11:12:37 UTC
author avatar
Vendor

ICS or ODI?

There are more products for integration.

Choose the one which is optimized for the backend system.

Finally, it depends on your budget.

You can try out products like Denodo, QlikView, NodeGraph

2017-11-23 08:10:45 UTC
author avatar
Vendor

Or you could look at HVR as an alternative, we provide Initial Load, CDC, Compare & Repair and reporting all out of the box. We have a Hub & Spoke Architecture, unlike peer to peer. Allowing many sources to be replicated once and delivered to multiple locations. We only need one access point to the cloud, again, unlike peer to peer. Easier to manage and considerable cheaper.

2017-11-23 07:51:12 UTC
author avatar
Consultant

If your use-case is near real time - e.g. transactions generated in ERP that need to update data in multiple other systems, then you might wish to go with ICS. If on the other hand you have large data sets (GB or TB) that is processed periodically, then you might use ODI. ICS is better suited to small, lightweight, frequent transactions while ODI would suit ETL for bulk data. Your use-case could have a combination of these as well..

2017-11-23 00:31:12 UTC
author avatar
Real User

Thanks much for the recommendations. Appreciate it! My thoughts are aligned as you are, but Oracle is recommending to go with ICS, as it is the future technology. I will take these inputs and move forward.

2017-11-22 17:20:27 UTC
author avatar
Vendor

I believe that ODI is appropriate for file processing and manipulation use cases. It provides most of the capabilities needed for such task through a large library of KMs / Knowledge modules. The required skill set for the ODI is mainly SQL developer which is easy to obtained easier than Oracle SOA suite or ICS developers.

2017-11-22 17:08:01 UTC
Find out what your peers are saying about Seeburger, MuleSoft, Matillion and others in Cloud Data Integration. Updated: May 2020.
419,214 professionals have used our research since 2012.