I’ve heard of this bimodal IT, what is it ?
Bimodal IT is a concept introduced by Gartner that seeks to assist large organisations who want to adopt the new coolness – Agile, CI/CD, DevOps – but are stuck with a legacy fleet of brownfields apps. The basic idea is to divide your development into two streams, Mode 1 which is traditional, waterfall development used mostly for your systems-of-record (SOR), and Mode 2 which is iterative and exploratory used for your systems-of-engagement (SOE). Since introducing the basic concept there have been further iterations that have considered a Mode 1.x which seeks to define a hybrid model which adopts some of the low-hanging fruit of automation without necessarily changing core platforms or practices.
OMG. That sounds amazing, so I get to be cool and I don’t have to deal with my tech debt?
Not so fast. Bimodal IT seeks to resolve a fundamental tension at the heart of the DevOps movement – changing core systems is hard and tends to be very expensive. On the face of it, Bimodal IT seems to be a way to have it’s cake and eat it too. Interact with more millenials and offer them fun mobile experiences, and you don’t have to touch that mainframe.
Jez Humble has provided a well-written and considered rebuttal of this panacea in his post The Flaw at the Heart of Bimodal IT. While he has a number of well considered objections the most serious challenge comes with the interlock problems introduced by this new two-speed change process. The reality is that innovation happens at all layers of the stack, and in many cases it is improvement to the systems-of-records which drives the cool new features that customers love. If you think about it, what modern web-based companies deliver is not just a fantastic mobile experience coupled with an evolving set of backend capabilities. Innovating top layer will help, but it will not be all you need to remain competitive.
Where does this leave us?
Another iteration touted by IBM and Accenture is Multi-Speed IT (discussed here on DevOps.com: Multi-speed IT vs Bimodal IT ). There is broad agreement that even the most prominent DevOps proponents that it doesn’t actually make sense to change everything, all at once, and so having a priority order of where you are investing in new practices is valid. Likewise, while COTS, Mainframe and Legacy are starting to benefit from newer practices the investment/return ratio will be different and you are going to hit unique challenges. For many companies this is a great opportunity to really re-examine the value and purpose of the legacy fleet, to consider the ongoing maintenance costs that have been accepted as just part of BAU to reconsider whether it’s time to make some sound strategic decisions in re-architecture.
- Gartner have both attained a lot of attention and received some criticism for their treatment of Bimodal IT. All of the major research firms are fighting to show they are still relevant in the DevOps landscape and so expect to see rapid iteration in this space.
- While we debate on how to apply the last batch of innovation to the Enterprise, the next round is already forming. Containers, Serverless, Read-only operating systems are all showing great promise. As the platforms settle, the next round of COTS apps will be delivered in new ways that are compatible with iterative development and can sit along-side in-house written apps.
- There are half-way houses worth exploring such as exploring in-memory caches, lambda architectures or big-data systems around your legacy SOR systems to both reduce pressure and increase capability. This doesn’t remove the need to increase agility in that space, but it may buy you more time.
- Keep up the conversation. While DevOps has proven its success many times over in the Enterprise environment, doing it at scale is hard and there is likely more than one answer on how to manage it well. Discuss with stakeholders at all levels to determine the path forward.
I would love to hear from you. What do you think of Bimodal IT?