There is a lot of talk these days from major IT research firms about how today’s enterprise CIOs need to embrace bimodal IT. One of the key ideas behind this bimodal IT concept is that CIOs need to build and support two different sets of IT infrastructure, one that is used for existing application environments (Mode 1) and one used for the rapid development and launch of lightweight, digital and mobile services (Mode 2).
These research firms go on to say that while traditional IT infrastructure should be modernized for Mode 1, where it will “run the business” using traditional commercial legacy software and hardware, CIOs should also build another set of infrastructure that uses open source software and commodity hardware to provide the DevOps team with the “startup” agility they need to develop and launch new applications quickly.
Unfortunately, this dual infrastructure approach rests on several false premises. The first is that startup DevOps teams are all using open source software, and that this is what enables agile application development.
The reality is far different. Most startup DevOps teams use a lot of paid software and services out of necessity because they don’t have the time or resources to customize and tie together a bunch of open source applications to meet their IT infrastructure needs. If they did spend the time building this infrastructure themselves they would never get their businesses off the ground.
Drawing on scores of on staff engineers and deep pockets, only the biggest of the big tech companies are building IT infrastructures that are based on open source and their own custom-built software.
The second false premise is that stitching all these open source applications is easy. It is not. What enterprises are likely to find is that while their IT team is busy trying to connect a bunch of open source software together, and its DevOps teams are figuring out and customizing all this open source software, their DevOps projects will not be moving forward. Instead, they will be sidetracked or delayed. It’s similar to buying versus building a house.
Packaged software provides a turnkey solution that is “move-in” ready at the time of purchase. Open source software is more like a collection of nails, screws and 2x4s. There’s still massive set up required to build it into anything resembling a DevOps infrastructure, let alone an enterprise class infrastructure.
The third -- and perhaps most significant false premise is that you need completely separate infrastructure for your Mode 1 and Mode 2 application environments.
Yes, it is true that traditional infrastructures can be rigid and inflexible, making it difficult for DevOps to use it to rapidly develop new digital and mobile services. However, it is also true that it is becoming increasingly difficult to support modern, mission-critical applications on this infrastructure as well.
The solution to shortcomings of traditional infrastructure should be to fix this infrastructure, and transform and modernize it so it can support both Mode 1 and Mode 2. In the end, the dual IT infrastructure approach potentially leaves CIOs with the worst of both worlds. They still have their complex, inflexible traditional IT infrastructure and in addition, they now have a new, completely separate piecemeal open source infrastructure that does not provide them with the security, scalability or control they need to deploy enterprise-level applications.
The obvious drawback behind this is the capital and operational cost of having to maintain their legacy infrastructure, which typically involves expensive maintenance and support contracts, plus the cost of having to build out a completely new infrastructure for the mode 2 applications.
Additionally, the complexity of having to manage two completely separate environments also becomes an inhibitor to growth because data governance, security and protection will now have separate technologies, methodologies and procedures. This could result in issues with compliance (DR for example) as well as potentially introducing vulnerabilities into the data center that could result in data breaches and other security risks.
If CIOs are seriously considering implementing bimodal IT -- as well as other modernization initiatives, such as creating a private cloud, increasing IT automation, and implementing big data and IoT programs – a more drastic change is needed. They need to embrace a Software-Based Infrastructure (SBI) model, where compute, network and storage are both software-defined and available on demand.
Will this change be disruptive? While it will disrupt current processes by introducing higher degrees of efficiency and accountability, it should not impact production operations. But it will provide enterprises with the agility, simplicity and scale of public cloud and the security, control and flexibility of their existing traditional IT infrastructure. And in doing so, it will create an IT infrastructure that can support traditional Mode 1 application environments and emerging Mode 2 environments that allow DevOps teams to quickly develop and launch new digital and mobile services.
Some might think that building an entirely new infrastructure is the only solution to the problem of launching and empowering DevOps teams. But the problem with this approach is it does not really address your DevOps growth challenges, nor will it allow IT organizations to support modern, enterprise-class application environments. If enterprises really want to adopt a bimodal IT approach, and support Mode 1 and Mode 2 requirements simultaneously, they should deploy software-defined platforms that are elastic, dynamic and can support all types of application environments.
(About the author: Jeff Alexander is vice president of engineering at Formation Data Systems).