So newfangled cloud technology is readily available on market. Now what?
Businesses across sectors are scurrying to modernize technology stacks, viewing the cloud as vendor-provided haven safe from sluggish, enterprise-operated data center technology.
It is important for technology business stakeholders to break the mentality that the cloud is the solution to enterprise technology concerns.
While people speak of the cloud in general ways, "going to cloud is not the goal. It is not success. It is not sufficient. It is necessary, but it is not sufficient," said Jeff Olson, VP and chief data officer of the College Board, in an interview with CIO Dive. But understanding why requires a bit of a history lesson.
In the 1990s and early 2000s, the dominant approach was to write "spaghetti code," where someone would write it up and down, tying it to other strands. This gave way to the mid-2000s rise of "lasagna code," according to Olson.
The new style of development centered on multiple layers, encompassing the full enterprise technology stack, from hardware and data centers up through the operating system and on to the web framework and mobile interface.
Each of those complex layers operate with separate costs, experts and requirements for security and patches.
"If you move to the cloud via lift and shift, and you're taking all of those layers and replicating them, you're kind of shoveling them from one location to another, which means you're keeping all that maintenance," said Olson.
The College Board, which overseas Advanced Placement testing and the SAT, spent years in "analysis paralysis" trying to figure out which cloud provider to commit to.
While it decided to center its architecture around one provider, Amazon Web Services, other organizations adopt multicloud solutions. This year, Rightscale found more than 80% of enterprises adopt a multicloud strategy.
To go with a single cloud provider, a certain amount of trust in the future is necessary, with faith that technology autonomy will remain in the future and market forces will ensure a single entity does not have too much pricing power, according to Olson.
Looking toward the stars
While the College Board used data to deliver services to students and educators, it is 118 years old and its software is heavily oriented around the 1990s and 2000s era. Highly monolithic applications mean there are a lot of technology interdependencies, with tight developed connections between software and the data.
That makes it "hard to make changes and hard to evolve without risk," Olson said. Partway through its modernization transition, the College Board is working to overhaul years worth of monolithic applications.
The College Board is not alone. Monolithic applications have a heavy presence in organizations that started 20-30 years ago. Companies are working to move to the cloud, but 65% still rely on enterprise owned and operated data centers, according to 2017 research from the Uptime Institute.
In the modern computing era, however, technology organizations can take advantage of services offered by cloud providers, rearchitecting software in a simpler way.
This technology simplification was one of the motivating factors in developing the North Stars, the College Board's seven guiding principles for its technology organization and key in its transition to more cloud-friendly operations. The North Stars make it easier to innovate without "bumping into other things," Olson said. At the same time, it breaks the organizational structure from matching the systems it maintains.
The College Board's North Stars
| Agile | The realization that there's a mirroring relationship between people and technology. |
| Core | Do the "fewest possible things and do them well." From there, Core means finding others in an organization "to do the things that you're not going to do well." |
| Microservices | Break technology apart into small machines with clean interfaces, surrounding them with tests. In doing that, each individual team can scale up or down as needed. |
| Cloud | Taking advantage of managed cloud services, which allows organizations to do less in a more secure and affordable way. |
| Data as Service | Move logic into databases that are open source, high-performing and where possible non-relational. College Board has a "minimalist thought about databases, like don't throw a big commercial data base at every problem." |
| Continuous Integration | Following the lines of continuous integration, continuous delivery (CICD), software teams can release more often, particularly when automated testing surround applications. |
| DevOps | Historically, organizations had one team write code and the other team operate it. But College Board is changing that, a principal gaining popularity in the enterprise. |
The goal for the College Board was to tackle the North Stars all at once, unsatisfied with partial adoption. The organization used teams associated with the original application monoliths to serve as "beachhead teams," rather than bringing in outside help.
It gave the teams ample resources so they were less encumbered, and provided coaching so they could tackle all seven objectives from the start.
The College Board said teams could operate faster, more efficiently and at a lower cost running previously-monolithic applications with regular changes and updates.
