by Myles F. Suer

How to fix the data integration mess

Aug 28, 2017
CIO Data Management IT Leadership

Once you've admitted there's a problem, it's time to figure out how to get started fixing it.

integration projects disasters 1
Credit: Thinkstock

Data integration represents a big issue for most enterprises especially as enterprise architecture becomes brittle and complex. CIOs and enterprise architects clearly believe they are responsible for architecture across all components of what is becoming an increasingly extended enterprise ecosystem. This includes on-premises, private cloud, IaaS, PaaS, SaaS applications, and storage. In fact, Paul Wright, CIO, Vice President of Information Technology at Accuride Corporation, said “if you’re not, then who is?”

When asked about the importance of leadership, Sharon Pitt, CIO of Binghamton University, said doing this right involves “supporting/facilitating use of/educating about architecture outside of central IT.” Clearly, building an integrated enterprise that can respond to the threat of digital disruption requires more than the IT organization. It requires the business and IT working together. Given this perspective, what principals should guide IT organizations as they build a multiple location/hybrid architecture? There are lots of ideas here, but I leave it to you to determine what’s most important to your organization.

  1. Start by making investment in strategy and an enterprise architecture that can respond to the future. Alongside this, make investments in personal, professional, organizational, and technical maturity.
  2. Unbundle the ERP, making trusted data available to all with rules to address security, permissions, etc.
  3. Educate business stakeholders on the impact of application sprawl and Shadow IT. Educate them as well on the cost, complexity, and security vulnerabilities that come from having a messy data architecture.
  4. Start by creating plumbing to legacy systems; as appropriate, rebuild these applications/data, and then move the pipes themselves. Typically, legacy data flows are a tangled mess.
  5. Consider what applications need to be refactored vs rewritten because moving legacy to the cloud can be very difficult if not impossible.
  6. Define cloud reference architectures with DevOps Continuous Integration and Continuous Development “in a box” to enable agile development for targeted platforms.
  7. Start unbundling services and data and then enable their use by stakeholders via microservices.
  8. Establish architectural standards, limit propriety design, limit overlap and multiple instances of a single business capability, use APIs, and establish data governance, trust, and business prioritization.
  9. Establish culture, skills, and of senior level management support for fixing the integration mess.
  10. Build on where things belong. Have an exit strategy before something goes live. Expect everyone to want to integrate with an application once it goes live.
  11. Document, govern, optimize and continuously improve your architecture.
  12. Plan for a change to existing business models. You can’t just do the same old business in digital disruption.
  13. Create flexibility, high levels of integration, and invest in doing the job correctly and completely during the build phase.

Parting remarks

There is clearly a lot to consider in the ideas described above, but taking on just a few can be helpful to CIOs building survivable organizations with enterprise architectures that work. Today, technology matters more than ever. Gary Roberts, CIO at Alfred University, says “we should feel responsible for specialized technologies…ultimately, when there are problems, we will be involved.” Clearly, CIOs and Enterprise Architects need to own the integration mess and create an operating model and enterprise architecture that enables their companies to achieve what Jeanne Ross at MIT CISR calls “greater strategic agility.”