There is an assumption that the increased use of automation will remove humans entirely from certain processes within an Enterprise. It is true that automation has taken on a large share of tasks that previously required hundreds or thousands of people, replacing repetitive manual tasks with technology, but Enterprises will always need human intervention. Sometimes the knowledge, decision making and thinking that are unique to humans are required to solve a particular problem, but the downside of this is that waiting for a human to take action can slow down a largely automated process and humans by nature are prone to make mistakes. In this article, we look at how service orchestration needs human intervention.
Human intervention does not imply a weakness, in fact quite the opposite is true if managed correctly. Subject matter experts are skilled at evaluating the uniqueness of critical business situations and rely on instinct and experience. The key is managing these human interactions efficiently within critical processes, presenting people with the best data available to support expedient decision making. In short, getting the right information to the right person at the right time, for them to make the right business decision.
Global organisations need speed to market, lean costs and accuracy. Removing human effort and human error through automating mundane, repetitive tasks such as repetitive configuration and administration will continue to be a key initial focus area. However, routing critical tasks between different teams and divisions, requiring human checks and approvals via email or phone is where the technical automation grinds to a halt and is bottlenecked with no visibility on status and with no visibility comes no accountability.
Three common examples where human intervention is necessary for automated processes:
1. DevOps sprint checkpoints
Even in the case of a DevOps technical release where everything is supposedly automated, there is a need for human intervention, such as recording key human checkpoints in a sprint cycle. DevOps tooling facilitates the automated deployment of code but doesn’t facilitate key human hand-off points that confirm checkpoints have been achieved. Validating and visualising that these human checkpoints have been achieved often remains a gap in tool sets, particularly where the DevOps processes straddle multiple processes, tools and teams. There are very few tools that come close to providing a single operational framework, whereas the best-in-class component tools are evolving all the time.
2. Exception management
In banks, Operations used to consist of thousands of people who had to check every single transaction that went through. As this became systematised over time, the amount of human intervention required dramatically decreased. Now humans don’t need to be involved at all in straight-through processing but will need to intervene if there is a missing piece of information such as missing client instructions. People are needed in these cases where judgement or knowledge is required that does not lend itself to auto repair.
3. Overnight batch scheduling
One of the highest-risk system areas in a bank is still overnight batch scheduling. Overnight batch scheduling involves processing the day’s data so the system can have a start-of-day position for the next day. The various systems involved in this process must perform their tasks in a certain order and if all goes well humans don’t have to get involved, but if a report is not sent in time or some other part of the process is out of sync, a person needs to be alerted so they can investigate the problem whilst trying to preserve the start of day critical path. This is a case where human judgement is needed to understand the best of what might be a set of undesirable options that limit the business impact in different ways.
In cases like this, there needs to be a system that can quickly alert the relevant people and allow them to confirm that the correct action has been taken. Service orchestration coordinates various automated tasks into an automatic end-to-end workflow, but it doesn’t address the areas where people are required to take actions that it is not yet possible to automate. Tooling for the human side as well as the technical side will reduce friction and reduce the amount of human effort required and the likelihood or human error causing problems.