In today’s world, every enterprise has a diverse set of applications and systems. Connecting those systems is often a requirement for business needs and is a challenge for their IT. This article emphasizes some of the most common problems in integrating systems.
1. Connecting to monolithic systems:
Let’s first agree that most of the enterprise will have monolithic systems and it is often not possible to get rid of those systems easily. This is because business heavily depends on such systems as they are mostly legacy systems and cannot be replaced for the sake of integration. A software system is called monolithic if it has a monolithic architecture. In monolithic architecture functionally distinguishable aspects are not architecturally separate components but are all interwoven. Due to such architecture, integration becomes a challenge. A systems integrator has to take care to identify and choose the best way of integrating with monolithic systems within the enterprise.
2. Different business units see data differently:
Use of existing enterprise data is critical to business success. Each business unit or business domain sees data differently because their operational responsibilities are not the same and they may be using different systems or applications. This is fine from an individual unit’s point of view, but a standard data model is required within the enterprise in order to integrate these systems. A systems integrator not only needs to satisfy the requirements of a business but needs to think beyond these and design an integration which has a standardised view of data within the enterprise. This is where a system integrator needs to focus more on modelling the data and ensuring a minimum canonical data model is achieved during integration. Yes! You read it right – “a minimum canonical data model”. The reason I use the term "a minimum canonical data model" is because many enterprise architects avoid using canonical data as agreeing upon it often involves multiple meetings and discussions to reach a consensus. These meetings are often avoided by individuals, but it cannot be denied that a minimum canonical data model should be chosen for integrating systems to gain a consistent view of data at the integration layer. This helps in avoiding point to point data modelling and allows other businesses or applications to connect to the systems using a service that implements canonical data modelling rather than having to integrate to individual systems directly in their native data model.
3. Do it fast or do it better:
When IT needs to deliver something to their business, they are often constrained with time and cost. Projects are often managed by non-technical managers who are business-centric and lack understanding on the best practices and architectural principles. Proper architectural planning, used to ensure quality often becomes difficult for IT due to these constraints.There are several tools available in the market that can speed up the project implementation and help achieve business requirements faster, but they often lack a good development pipeline and use of the best architectural principles. A system integrator must find a balance and ensure that the integration solution is not just about meeting the functional requirements of today, but also potential future and non-functional requirements such as performance, scalability, re-usability and maintenance.
4. Choosing the right tool for integration:
There are plenty of tools available in the market that promise to deliver the best integration experience. Choosing the right tool for an enterprise is a challenge. In modern connectivity, an integration tool must almost certainly support hybrid integration. Hybrid integration allows on premise applications to seamlessly integrate with cloud based applications. It bridges the network between your existing enterprise and SaaS/PaaS, B2C, B2B, BYOD and Big Data. Hybrid integration is desired as enterprises may not want to go completely on cloud due to depending on legacy systems, regulatory compliance issues, and concerns around security and privacy of some sensitive data. Hybrid integration tools allow addressing these gaps and helps business leverage new applications on top of their existing systems using a service or API layer.
5. The major challenge with integration is integration:
Yes I’m not kidding. "When you go to IBM, you get an integrated IBM solution. When you go to HP, they lead with a HP solution. But when you say, 'I already have Dell servers, and I'd like to add HP storage,' that can become problematic, and that's where the integration challenge comes from," says Roger Channing, Chief Technology Officer at Cogent. In general, an enterprise chooses a stack of products from a particular vendor and only later do they identify that the stack is either very complex to maintain or does not satisfy their needs. Choosing right set of products for your business is very important. The product chosen should allow integration with other systems or products either via a web service layer or using libraries. This will help enterprises avoid another monolith and designing and implementing integration solutions for their current or future products.
If you would like to find out more about how MuleSoft Systems Integration could help you make the most out of your current infrastructure while enabling you to open your digital horizons, do give us a call at +44 (0)203 475 7980 or email us at firstname.lastname@example.org.
Other useful links: