Recent stories in the press show how migrating applications and their associated data to the cloud isn’t as simple as lift and shift. Sure, the marketing phrase makes the effort sound easy: what can be simpler than taking ‘process A’ from one location (ie on premises or in a data centre) to another (ie cloud service)? The problem might be that too many vendors use the ‘lift and shift’ label for what is in effect a brand-new or re-implementation.
The first article is this: DexCom Cites Switch Outage, from the WSJ, Dec 18 2019 (link here is with MedCity News). The second concerned a local issue at Marta, Atlanta’s city consumer rail network. There was a report that cloud migration of a billing system linking credit card providers to the ticketing machine network had gone down during the cloud migration.
In both cases what seems to have failed is not the process being lifted or shifted. It seems to be the integration between the lifted/shifted process and others connected to it. More insidious here is this: it’s not just the technical integration but it is more likely the semantics in the data shared in the integration that are inconsistent and that lead to the failure. In other words, the integration ‘works’: data is physically and so successfully exchanged between app/process A (migrated) and app/process B (wherever it is/was).
The issue is the two apps/processes interpret the data differently. Thus, failure takes place even though technology seems to work. In this sense, IT is not one thing: it stands for information and technology. Technically a lot of integration works; semantically information is garbled. For a boomer (ok boomer) like me (according to my youngest son, though technically I am not a boomer) this takes me back to my EDI days.
We called out the pending nightmare here some years ago – see Beyond the Gathering Storm: Governing Data and Analytics in the Cloud. A few early clients called out the issue though the inquires were not pleasurable affairs. I still remember clearly the sounds of an exacerbated bank CIO who was literally being forced to open up his firms’ check book by a large software vendor/systems integrator who had just completed a cloud migration for a large transaction system. The customer data being exchanged between the new cloud app and the original systems was garbled; even though technically all systems were integrated. Data quality, master data management and data governance were an afterthought. The CIO was being told that he needed to spend more, much more, money to implement “real time cloud integration!” What madness that call was.
As more and more apps are moving to the cloud, and disparate clouds for that matter, you need to step up your integration strategy and recognise that it is not really about technology: that’s the easy part. An API strategy is easy: governing semantics in data across apps connected with those APIs is harder.
*This article is reprinted from the Gartner Blog Network with permission.