Bimodal IT is only harmful when oversimplified


Kurt Marko wades into the bimodal IT debate with the alternative view

Business man clutching head superimposed on two-lane highway © .shock -’ve been asked and written about bimodal IT numerous times over the last few months and have developed a nuanced view of the concept that doesn’t comport with the dire warnings summarized in an earlier diginomica analysis, Gartner’s bimodal IT considered harmful.

Although the bimodal concept can be polarizing, I believe much of the blowback originates from assumptions made due to an unfortunate choice of name, reflexive distaste for analyst buzzwords and particularly the term’s originator, the analyst firm so many love to hate. A common construction takes bimodal to mean bipolar, with IT segregated into two separate, but unequal entities: Mode 1 where all the stuffy IT old-timers live out their days caring for decaying databases and molding mainframes, versus Mode 2 where all the cool kids play with the latest toys and work unshackled from IT bureaucracy and processes. If that’s your view, bimodal is a recipe for disaster: a warring, dysfunctional IT organization.

If one subscribes to the us-versus-them characterization, I fully agree with ActiveState’s CEO cited in the earlier analysis that

…one can expect many companies to experience huge conflict as the two camps engage in pitched battles for influence, resources, and power.

However, I contend that any IT executive that implements bimodal as a caste system with implied winners and losers misses the point and is guilty of gross mismanagement. Indeed, as I wrote last summer, requirements-based segmentation has been going on since the dawn of IT:

Whether you call it legacy versus emergent systems, Brownfield versus Greenfield deployments or sustaining versus disruptive technologies, the dichotomy between old and new or maintenance and development has been around since the dawn of IT. Each category has always required a different set of investment, management and governance techniques. The difference now is the pace at which new products are developed and refined and a concomitant decrease in useful half-life of mature services.

A key problem in much of the bimodal debate is that it overemphasizes the importance of Mode 2 development and minimizes the innovation, rejuvenation and reinvestment required in Mode 1 systems to maintain competitiveness in the era of digital business. As I put it here, the reductionist view of bimodal IT

…understates the amount of innovation and service improvement that needs to happen in Mode 1, business critical systems, creates a false dichotomy concerning cloud usage within IT and romanticizes the nature of Mode 2 work.

The flaw is in assuming that Mode 1 systems are on life-support instead of in need of some life-saving surgery and facelifts. Indeed, I completely subscribe to the view expressed in the previously-cited quote from analyst Jason Bloomberg,

What many organizations are finding is that for digital transformation to be successful, it must be end-to-end — with customers at one end and systems of record at the other. Traditional IT, of course, remains responsible for those systems of record.

Yet the need for IT transformation doesn’t invalidate the bimodal concept, it underscores the need to do things differently since the pace of change, degree of risk and tolerance for mistakes in these Mode 1 “systems of record” cannot be so high that it jeopardizes critical business operations. These require stable, reliable, highly available applications using mature and well-tested systems, however they can’t be static and moribund. A key tenet of bimodal IT is what Gartner calls renovating the core. To me, this means bolstering mission critical infrastructure and applications with new technologies like distributed, scale out, microservices-based designs, virtualized, containerized cloud stacks and use of public cloud services whether for bursting, DR or implementing new features. It’s important work that requires innovative engineering and IT resources, but with deliberation, security and risk control required of mission-critical business services.

In contrast, Mode 2 is the place for IT experimentation and risk-taking. As I wrote earlier,

Mode 2 provides the structure, or lack thereof, for IT and developers to learn, inculcate and perfect the behaviors and technologies required to attack fast-changing prospects for digital business through new applications and services.

Here the emphasis is on new products and services in often dynamic markets with unknown odds of success. Mode 2 provides a structure to perfect processes in agile development, continuous delivery and rapid, data-driven customer feedback for both application developers and infrastructure architects. In this wave of IT innovation, whether you call it digital business, 3rd Platform or just today’s competitive reality, both the business opportunities and customer tastes are uncertain, fickle and often fleeting. The goal in Mode 2 is to maximize IT’s ability to create, adapt and react while minimizing the cost of failure.

Another way the bimodal debate sometimes mischaracterizes the desired organizational structure is by erecting walls between the two parts of IT. Again, this is understandable given the name and a superficial look at the concept, however it’s a misreading of bimodal in my view. Instead, I see successful Mode 2 behaviors migrating throughout IT over time, an osmosis that brings increased dynamism and adaptability to all of IT.

My take

Perhaps Phil Wainewright is correct that enterprises trying an all-or-nothing IT transformation will find it easier than feared, but count me skeptical, particularly for large organizations with hundreds of legacy systems feeding dozens of business critical processes. To borrow my earlier metaphor, bimodal IT is old wine in new bottles, however unlike the skunk works of old, Mode 2 can’t be isolated from the rest of IT, just insulated from the innovation-sapping, risk-averse bureaucracy. Done right, bimodal IT should inject fresh thinking and faster, more efficient processes without fracturing the organization into new-versus-old, us-versus-them tribes, where Mode 2 plays the role of catalyst in digital business transformation.

Image credit - Business man clutching head superimposed on two-lane highway © .shock -

    Comments are closed.

    1. My longview based on Kurt’s, Phil’s (and Den’s weekend rant on Logfiles).

      Architecturally bimodal is unavoidable for large organisations. Mainly because early SaaS ERP fashioned on class multi-tenancy architectures is self-serving for software companies and not refective of newer micro services cloud based functionlity capable of being written now (but generally not in ERP software companies).

      The iPhone and Android ecosystems show the post PC era architecture is changing to low latency high user experience sync clients coupled with cloud services. Den has even covered architectural break through (but Jut domain know-how is logfiles not ERP). Google’s is more illustrative.

      Either someone like Workday will need to publish reference architecture to build nextgen apps. Google published BigTable and Amazon the DynamoDB papers setting off the NoSQL database boom.

      Perhaps we need to wait for Metavine / zapier to break on through the myths of current SaaS ERP. Where functionality purchases are driven by finance and sales types and not by users generating revenues. Users are bringing their iPhones / Androids to work to get work done on consumer tech (not slacking off on Slack).

      Coda, this shift in architecture and lack of direction for ERP is why the PE boys continue to scoop up. Sage Group Plc. TBD.

    2. Phil Wainewright says:

      @Kurt if you’re going to redefine bimodal IT as requirements-based segmentation within the context of rapid digital transformation then I’ll have no argument with it. The reason I called the approach harmful in my original article is that people are citing bimodal IT as a justification for postponing necessary action to bring legacy systems of record up-to-date with the needs of today’s connected digital business environment. It reinforces the, if it ain’t broke, don’t fix it, school of thought that has been holding back meaningful innovation in much of enterprise IT for decades.

      I think that if an organization agrees that all its IT teams have to work collaboratively to address today’s business demands then what they end up doing will involve different systems and applications transforming at different speeds, but no one will find Gartner’s bimodal IT a useful framework for describing that phased transformation.

      Clive interestingly takes the argument in a different direction, stating that many of today’s SaaS vendors are already legacy platforms and so there’s always an element of bimodality in an organization’s IT architecture. That’s a fair point, but I wouldn’t want to elevate it into something that I welcome, merely as reality that has to be acknowledged and dealt with. It’s more of a bug than a feature.

    3. Tony Price says:

      “osmosis that brings increased dynamism and adaptability to all of IT” here here!!! At last some some common sence and thank you for this.

      I totally agree that Bimodal is so often interpreted with negative conitations. The “old world”, “legacy IT”, “yesterdays technology” are all words I see being used as if this mode is something bad ……. but todays technology is tomorrows legacy and it will continue to be that way. So dealing with all types of IT holistically is just the reality of the world we live in.

      Thanks for this article!