Application Development

ERP: Where DevOps Breaks Down

DevOps is no longer something that you hear about only at Netflix, Facebook, and Amazon. In fact, most Global 2000 companies have already created a DevOps initiative. But the reality is that most CIOs have yet to fully embrace DevOps.

Brett Stevens

Sep 12, 2016

A CIO's Dilemma

DevOps is no longer something that you hear about only at Netflix, Facebook, and Amazon. In fact, most Global 2000 companies have already created a DevOps initiative. But the reality is that most CIOs have yet to fully embrace DevOps.

One reason that DevOps hasn't gained complete traction is because CIOs are balancing many different priorities. According to Ralph Loura, previously CIO at HP Enterprise, many CIOs are juggling the technical debt associated with legacy systems along with projects employing big data analytics, open source, PaaS, and a myriad other emerging technologies. (For Loura, there was also the matter of a Fortune 10 company splitting in two.)

Then, there's the obvious challenge of creating a new IT culture and processes around DevOps. It's one thing for the next generation of apps in the cloud, but there's a giant pink elephant in the room: every company's ERP, a complex legacy system with hundreds of interconnected applications. The number of articles and blog posts titled "DevOps for Oracle EBS" or "DevOps for SAP" are scarce, and for good reason. But how can a company ever achieve full-scale DevOps until these systems are addressed?

The Trouble with DevOps and ERP

Perhaps the biggest impediment to DevOps for ERP is what PwC calls the "agility versus stability paradox". The high change rate that comes with DevOps is difficult to swallow for teams maintaining the "sacred cow" that is ERP: that is, mission-critical systems that rank at the top of the list in terms of costs of application downtime. The shift in rhythm required for DevOps--such as deploying code changes daily versus weekly, monthly, or even quarterly--better maps to systems of innovation and differentiation than systems of record. That's why Gartner proposes "bimodal IT" and McKinsey a "two-speed IT architecture" to draw the line of bifurcation between apps based on their max RPMs.

Beyond a shift in IT's view towards introducing changes to the business faster, there are a larger set of challenges that CIOs must address around the scalability of process and infrastructure for their ERP systems. Creating a development environment for Oracle E-Business Suite (EBS), for instance, might take six weeks of manual coordination across five teams (such as the DBA, storage, systems, network, and VMware teams) at a large IT organization. That's not exactly scalable when you have to create 5, 10, or even 15 of those environments and keep them up-to-date. Not to mention, ERP systems are often some of the heaviest users of storage and server resources in an organization.

Automation is another pillar of DevOps. The end goal is for developers to be able to test their code on realistic, production environments without involvement from operations teams, enabling them to discover and remediate defects early in the software development lifecycle (SDLC). Compare that goal to the HP LaserJet firmware team before their DevOps transformation. From the time a developer checked in code, a full integration test required eight full weeks to determine whether their code change worked or not. The reality is that many components must be automated across the various stages of the SDLC for DevOps to work.

In the world of ERP, many projects get delayed. Business units become frustrated that their internal IT team is slow to respond, and often, not error-free. But what if there was a way to have both agility and stability?

The First Step to DevOps = Data

There are many tools out there to automate the build, release, and deployment stages of the SDLC (see the DevOps Periodic Table by Xebia Labs below). But that also means that there are many interchanging parts with an ambiguous ROI.

Data is the fastest, easiest, and biggest bottleneck to address when it comes to DevOps and ERP. Take Delphix, for instance, which is software that automates the provisioning of a full-stack EBS environment--including the Oracle database, dbTier, and appsTier--and shrinks the end-to-end process down to minutes. Also in Delphix's technology recipe is the ability to reduce the storage required to support multi-terabyte EBS project environments by up to 90%. That means Delphix has the twofold benefit of putting you on the course to DevOps while also providing large and immediate cost savings.

Unlocking data trapped in legacy systems is critical because it impacts how quickly IT can deliver features to the business. One Delphix customer, for instance, has reduced the time to provision an Oracle EBS environment from 18 hours to just 30 minutes. Think about that in the context of a global company with over 450 products and more than 3 million distributors in over 100 countries and territories, each with geo-specific business process requirements. A lengthy refresh process prevented developers from quickly testing and releasing code changes using the latest production data, and ultimately, impacted their ability to quickly roll out new EBS features and enhancements to each country and territory in support of rapid company growth.

A CIO at a corporate tax software company explains how they are similarly using Delphix for their ERP systems. "We increased the velocity of our internal testing cycles--not only how quickly they went, but also the number we could do in a period of time. As a result, we can introduce capabilities to our business much faster than we were able to before."

Taking it one step further along the continuum of DevOps, self-service breaks down silos between systems admins, developers, testers, and release and operations teams. According to one IT administrator at a Fortune 100 High Technology firm in a recent TechValidate survey of Delphix customers, "we have reduced the need to coordinate efforts across storage, solution architecture, and DBAs. Instead, it is just one click to refresh a database." Or here's another: "It takes us a couple of minutes versus days to refresh multi-terabyte databases for development and testing of our business-critical applications. And we can refresh without errors or IT staff assistance."

One testimonial that particularly stands out is by an IT Manager at a Global 500 Banking Company: "Delphix is a game-changer. We are already benefiting from the reduced workload on operations through end user self-service. We are leveraging automation with Delphix to further drive down operational overhead and are starting to see even greater savings with more to come."

These testimonials indicate that you CAN have DevOps for your ERP, but it is a journey with many obstacles. If you're looking for the path of least resistance, data is the first component you should look to automate. It will make your ERP systems both more agile and stable at the same time.