Notes | The Banking Industry Underestimates Costs of Cloud Migrations
DOI: 10.1109/CB149978.2020.00039 | Source: IEEE Conference of Business Informatics | Authors: Fadime Kaya, Martin van den Berg, Roel Wieringa, & Marc Makkes
Starting the evening with a 2020 paper on the banking industry underestimating cloud migrations costs. This one was longer than I expected wen I moved it into my to read folder.
The authors chose the banking industry because it was one of the first to adopt IT and banking products are ephemeral and more easily translate into digital services. (302). The researchers were unable to do quantitative research because IT related costs are usually tracked as one general cost instead of being subdivided (302); however, the authors were able to organize costs discussed in the interviews into nine discrete categories. They find that banks underestimate managing dependencies, the cost of re-architecting applications, hiring external contractors, and legislation. They conclude by noting that banks underestimate the costs and that the banks that participated in the study expected migration costs to further increase and become even more volatile. Furthermore, that because cloud computing is still a nascent technology with many unknowns, they caution IT managers to not approach cloud migration as a typical IT project.
Quotes & Observations
“Cloud services can rarely be used immediately as it requires customization depending on business requirements, and positioning within the architecture.” (301)
“A stakeholder analysis identified significatn disadvantages to the benefits, long term costs were volatile and to a certain extent vendor lock0in occurred to the diffusion of control and dependency on the cloud vendor” (301)
“Cloud vendors pledge lower costs with their “pay-per-use” utility slogan. However fail to distinguish which components are standard and which components require re-work for which a company needs to pay for.” (301)
“Data storage, outsourcing data of customers and privacy sensitivity are underestimated problems” (308). In addition to this, without overarching legislation many countries have their own set of rules.
“Most banks do not manage cloud migration costs separately, which does not contribute to understanding and predicting cloud migration costs.” (308)
Five Migration strategies
Rehost: “The application is running in and Infrastructure-as-a-Service (IaaS) environment. Cloud is basically approached as a data center. The benefits from cloud will not be reaped with this strategy and it is difficult to scale” (301)
Refactor: “Enables developers to re-use langauges, code, and containers. As a result the application is “cloud native”. This might pose framework lock-in and is not loosely coupled.” (302)
Revise: “The application will be altered extensively prior to the cloud migration. The current code will be modified. This is the case for legacy applications/end of life applications which requrie revision. It is an expensive strategy and time consuming.” (302)
Rebuild: “The existing code will not be used and is discarded. The application is usually re-build[sic] on a Platform-as-a-Service (PaaS) stack. Reuilding ensures that capabilities and features of the cloud vendor can be taken into account to gain the full benefit of cloud.” (302)
Replace: “The application is discontinued and replaced with a Software-as-a-Service (SaaS) solution. Vendor lock in of the application can be a risk, but is an efficient solution in terms of time and costs.” (302)
The researchers identified nine cost categories.
Eight out of ten banks underbudgeted for managing dependencies during cloud migrations (304). With each dependency the institution must decide if they will migrate all of the services that are interlinked. In this case study there were also instances of the banks not fully comprehending their IT landscape (302).
Seven out of ten banks underbudgeted on complying with existing regulation. Regulations are often behind the adoption of new technology and when the banks went to implement a new solution they were often required to check with individuals from each locality. Additionally, at least one interviewee expressed intense time and money costs to getting answers to questions like “where and how is the data stored” from the cloud provider (304).
Support From Other Departments
Five out of ten banks underbudgeted. One bank had all of their requests filled on a per request basis and thus had no idea on costs. Another slowed the adoption because they lacked the expertise to go through. (305) Security, compliance, and risk are the main departments that will be needed during a cloud migration; however, there are likely other stakeholders across an institution who should be involved and have a relationship with IT departments. According to the authors, the banks that did this, had “increased agility and effectiveness” (305) with the cloud migration.
Five out of ten banks underbudgeted. The authors note that, “during a cloud migration banks are not fully aligned yet with the possibilities offered by the cloud vendor” (305). Sometimes it’s not possible to re-architect in the cloud due to requirements or the differences between the existing technology and the cloud services themselves (mainframes are still online running Cobol for a reason).
None of the banks spent less than budget on external contractors and 5 of them over spent (306). External contractors are often hired because the internal talent often does not have the knowledge. The other factors that lead to surprising external contractor costs are: a lack of preparation from project management, changing business requirements, and IT employees being told it will happen sooner than is feasible. This latter point can lead to employees moving on to greener pastures and institutions should have a retention plan, else they can end up with new hires that do not know the internal system (306).
Less banks overspent on this category (4/10). The reason in every case was too much optimism regarding the amount of internal knowledge regarding cloud migrations. Those involved with the migrations noted that the cloud presents a steep learning curve (306).
Four out of ten banks overspent in this regard. Some under spending was due to a lack of interest from internal staff; however, firms that engaged in training discovered that the knowledge gap was worse than when they began. Whether or not the bank spends money on training resources and if they promote the available training also has an impact on learning outcomes (307). We can infer from this that sustained knowledge gaps will increase cloud costs.
Cloud Migration Core Team
This is an area where only three out of ten banks underbudgeted. The primary reason for increased cost is the complexity of a cloud migration. These banks also noted that because experience with cloud computing is very low in the industry, it requires more personnel than initially imagined (307).
Third Party Services
The authors note that third party services differ from external contractors because the service may manage a process end-to-end. One interviewee mentioned that it was simply too costly to migrate a mainframe application and the return on investment would be negative. Another bank complained of lock-in from an existing vendor. A third party service managed their on-premise environment and the vendor was simply not interested in working on helping them migrate to public cloud (307). With third party services there always exists the risk of vendor lock-in.
The Technology Business Management Framework (TBM) might be useful to look into as it is a method to align IT and Finance departs and increase the value of IT spending; however, it also is very likely not relevant to where I want to go with this project.
For future projects I am interested in their data collection method (302-303).
References to Read
“Risks in Enterprise Cloud Computing: the perspective of its experts,” Journal of Computer Information Systems
“Unlocking Sustained Business Value from IT investments, ” California Management Review
“To Move Or Not To Move: The Economics of Cloud Computing”, Hot Cloud
“Critical Analysis of Vendor Lock-In and Its Impact on Cloud Computing Migration,” Journal of Cloud Computing