Challenges of DevOps at Scale
Organizations today are dealing with a large number of applications, devices, and data introduced by cloud, mobile, big data, and the Internet of Things (IoT). With the advent of automation and continuous everything, the situation is likely to get more challenging. Enterprises try putting more stress on having a clear DevOps strategy for the workflow to move smoothly. But, as DevOps is a result of efficient synchronisation between developers and IT team, making it work at scale can be met with many glitches. Here are a few common issues generally faced by teams in the DevOps workflow:
- Too many interruptions
While working with two or more teams, there are bound to be interruptions. Like in every team, IT operations will have planned and unplanned work coexists by design. For them, engineering tasks like analyzing new technologies to deliver new platforms, building suitable environments for the applications, and working on scaling and performance are permanent. But, along with them, come the constant interruptions due to incidents that require immediate attention like security events, customer requests, and sudden outages. All such interruptions are time-sensitive and require the immediate attention of the team. It might be impossible to have a separate team just for addressing such interruptions due to the uneven distribution of knowledge and skills as well as limited resources.
Thus, every team ends up juggling all types of work at one point or another. This, in turn, delays the tasks related to app delivery and extends deadlines.
- Silos destroying workflow
Most organizations trying to implement DevOps workflow have teams working in silos. Silos create segregations based on work specialization and these divisions end up becoming barriers eventually. In theory, silos appear to be a well-organized system where each team handles tasks according to their capabilities and passes on the rest to other teams, i.e.: creating request queues. But, in practice, these request queues can at times, be the death of steady workflows. As discussed before, interruptions are unavoidable in every team. The delay effect of interruptions is cumulative in linear workflows.
- Longer waiting time as tasks are codependent
Each team in silos are dependent on other teams to complete tasks. Also, all teams have separate task priorities. Considering this, each team loses time while waiting for the other team to do something for them and vice versa. So, unless a particular team is an expert in switching context and accomplishing tasks assigned to the other team for completing workflows without delays, there is surely going to be a lot of delay due to this time spent waiting around. Delays slow down the feedback loop, and a slow feedback loop leads to more errors. Delays also increase the likelihood of a change in the context of the work, causing further issues. In short, even small delays can compound into big problems.
Enabling Smooth DevOps at Scale
These glitches can be difficult to resolve as there is no straight-forward solution for any. That is why organizations are leaning towards pursuing self-service DevOps. But, for enabling manual self-service DevOps, one either needs to be well-versed with REST, ESB, SOA, SOAP and all the needed context switching or else completely rely on IT teams to build each integration beforehand. Either of the options seems very tough to achieve but can work wonders for enabling continuous delivery. But, considering the gaps in the workflow, there has to be a system in place that can enable self-service app delivery.
Imagine if there could be a self-service kiosk for the purpose, just like a bank Automatic Teller Machine. There are 2 contributing parties in the operation of the ATM machine — the bank and the customer. Customers can perform simple tasks like withdrawing or depositing money all by themselves without contacting the bank executives. Similarly, if there could be a kiosk-like system where the IT operators could predefine and build all the required app delivery artifacts like preconfigured environment profiles, containerized app stacks, choice of target infrastructure, etc., developers can add their needed specifications to the suitable integration and deploy apps on their own accord without having to contact the IT teams.
Consider a modern application with a microservices architecture with X number of services. Chances are there will be more than one microservice for each service. Now, for building each microservice, there will be a separate team of developers working on it and each team obviously will have a different work pipeline. The output of each pipeline from the development team will be dumped on IT’s table leading to a classic model of silos workflow. But, if there was a way of letting the IT team have an independent work pipeline, wouldn’t that simplify it all?
This is what self-service DevOps can offer. Self-service DevOps is a way to enable developers to deploy applications on demand all by themselves as the IT team has done the needful beforehand at its own pace. This way, there are minimal interruptions and delays in app delivery, thus achieving a state of continuous delivery.
In short, enabling self-service DevOps brings forth the following benefits:
- Drastically cut cycle times
- Accelerate releases
- Reduce application backlogs and other bottlenecks
- Reduce operational costs and increase operational efficiency
The idea of having a virtual self-service kiosk for software delivery can actually be implemented by either having a process in place or having an app delivery platform that can offer such functionalities. A well-built app delivery platform can streamline the CI/CD pipeline for an application of any scale. The idea is to have a predefined app delivery process in place that can enable any enterprise to unlock the benefits of DevOps with ease.