Skip to content

Don’t Implement a Data Lake Sandbox Without a Strategy and Roadmap

, | March 10, 2017 | By

By Jeff Kish – 

Today, many Big Data projects begin as technical sandbox projects, intended to help IT teams understand the viability of the technology within their data ecosystem. The technical aspects of the project go just fine, but determining real business value from their investment in the project? Not always the case. In fact, further investment requires an understanding of business value and capabilities that will determine where data will reside and how it will be used in the modern data architecture.

Organizations will often talk themselves out of planning and strategizing for their Big Data project. “We don’t need a strategy,” or “we’ve done strategies and roadmaps in the past and they always change.” This will lead to future issues and sometimes failure — or at the very least, causing frustration for users. The very nature of a Big Data project requires an understanding of where is the best location for data within the modern data architecture

No Strategy Is Not A Strategy

There’s a famous quote by the military strategist Helmuth von Moltke: “No battle plan survives contact with the enemy.” But the great Prussian military leader was not advocating going to battle without a strategy. Instead, he was emphasizing the necessity of flexibility in adapting strategies to changing circumstances.

Organizations have strategies for achieving goals, and so should individual projects — including the implementation of a Big Data environment.

Building a Roadmap

Why do companies jump into Big Data projects without strategizing and planning? It’s likely because they learn of the success others are having with Big Data projects — maybe even their direct competitors. The potential of reduced time to value generates a fire of enthusiasm for getting a project of their own underway…right now!

In such cases, the enthusiasm and competition are powerful, yet can be misdirected.

The need to move quickly is certainly understandable; many companies have already reaped big benefits from Big Data analytics. And the age of Big Data is here.

But if you’re planning to jump into that Big Data project or even a data sandbox implementation, take some time to channel that enthusiasm into planning that leap — or, to put it differently, looking before you leap. You can do that by building a roadmap to success.

It’s important that the roadmap is based on a business imperative, but it’s also important that the roadmap doesn’t just focus on a specific technological discipline. When building your roadmap, it is critical to focus on business capabilities (use cases) and to exercise caution in basing it solely on a technology

Successful organizations understand that investing in a change of direction with Big Data is not just a technical project, but rather a transformation of their information environment. Starting out with a specific plan that clearly sets your direction will improve your chance for success.

Consider the following for guiding the formulation of your plan:

  1. Define the Business Value: Do you have clearly defined benefits and values — including use cases, where applicable — that your organization will derive from the implementation? Defining business value should focus not only on the initial implementation but also upon follow-on phases.
  2. Address Leadership Positions: Many first forays into Big Data projects end up drifting due to key leadership positions not being filled, or failure to put in place clear business and IT leadership structures charged with guiding the project. Today, many organizations have established a Chief Data Officer as a key leadership position. In addition, a Business Analyst and Data Scientist should be designated that truly understands the business needs that the project will seek to fulfill.
  3. Define Architectural Vision: Firm-up the vision of what the finished product will look like, and how it will serve end users’ needs. Use that vision in conjunction with your business value drivers to determine which architectural capabilities are needed sooner than others, in addition to natural technical dependencies.
  4. Formulate Your Operating Model: Your organization needs to be planned. In addition, data governance organization and processes need to be finalized.
  5. Establish a Roadmap: The final product of this 5-step process will be a roadmap that serves as a project guide, including the order of operations and the listing/defining of capabilities that will be needed.

Failure to take these into consideration will likely lead to many problems, including:

  • Conflicting expectations about what is being delivered in a specific timeframe
  • Questions about the business value of the project
  • Inadequate staffing issues in developing, supporting, and managing the project

Adjusting Course

No matter the time and effort invested in creating a roadmap for your project, changes to that roadmap are inevitable. And that’s a good thing because changing the roadmap as needed helps assure that you’ll stay on course toward your goal.

In fact, I recommend that you not wait for organic changes to your roadmap to occur. Instead, review the roadmap periodically to look for needed tweaks; do this at least twice per year.

When changes to the roadmap are needed, don’t waste time in blaming the original architects of the roadmap. Changes are inevitable, and nobody can foresee all possible eventualities. Just implement changes as needed, and continue moving ahead.

Plan Your Success

It’s such an overused saying: failing to plan is planning to fail, or at least stall. But it’s so overused because it’s so true. It applies to wars. It applies to long-term corporate goal setting. It certainly applies to that data lake project that you’re just now thinking about implementing.

And it’s a great way to assure that your data lake sandbox project doesn’t morph into a patch of quicksand — into which your Big Data hopes sink into oblivion.