devopsy.kinsta.cloud

Tag: platforms

What Do DevOps Engineers Think of the Current State of DevOps
Over the years, DevOps has been called many things: A culture, a practice, a mindset. Today, it’s increasingly framed as a platform, coalescing with platform engineering, cloud native maturity and next‑generation AI. But what do DevOps engineers themselves think about where we stand in 2025? Having spoken to dozens of professionals, combed through the latest surveys and observed trends closely, here’s a look at what the trenches are telling us.

Platform Engineering’s Domination: A Double‑Edged Sword

Platform engineering has moved from buzzword to battleground. According to a Red Hat survey conducted in September‑October 2024, nearly half of organizations say generative AI is a core component of their platform engineering strategy. And platform engineering maturity strongly correlates with success, particularly when organizations treat their internal platforms as products with measurable KPIs. (Red Hat)

DevOps engineers commonly report relief in daily toil — IDPs and golden paths reduce friction and enable faster delivery. But complaints persist: building and maintaining a platform can be costly, inflexible, and siloed unless it’s scoped carefully. One recent report emphasized that challenges like workflow integration (37%), security risks (37%), and skills gaps (34%) remain barriers even for advanced organizations. (Red Hat)

In short: Engineers appreciate platform engineering — so long as it’s well built, developer‑centric and not a one‑size‑fits‑all. Otherwise, it risks becoming another source of friction instead of productivity.

AI’s Two Faces: Agentic and Generative in the Trenches

Generative AI is no longer a novelty — it’s everywhere in DevOps workflows. Surveys show 83% of organizations plan to increase spending on AI code generation, and 76% on agentic AI in the next 12–18 months. (Red Hat, devopsy.kinsta.cloud) Tooling that automates documentation (76%), generates code (74%) and provides intelligent code suggestions (59%) is already standard fare in many shops. (Red Hat)

But Ai’s influence goes deeper. TechRadar recently identified three generations of AI coding tools: From Copilot‑style completions, to in‑IDE agents, to full SDLC‑integrated agentic systems. As of mid‑2025, tools like GitHub Copilot, DevOps agents and Zencoder Zen are automating entire pipelines — almost like AI co‑engineers. (TechRadar)

DevOps engineers I speak with are cautiously optimistic. They see agentic AI relieving routine toil: Spinning up IaC, reviewing compliance, optimizing cloud costs. Platforms like CloudBees Unify now connect toolchains under AI‑enhanced control planes. (Wikipedia) At the same time, many practitioners note that AI is still brittle — it works, but only when supervised, prompted carefully and monitored.

The rise of “vibe coding” — a nearly free‑flow, conversational style where engineers guide LLMs rather than write every line — is increasingly common in startups, and sneaking into enterprise prototypes. (Wikipedia)

DevOps Tools Maturing into Platforms

The toolchain is consolidating. CI/CD, monitoring, compliance, security and cloud provisioning tools are increasingly bundled or bridged in platform layers. devopsy.kinsta.cloud’s coverage tracks this trend: It’s no longer about separate pipelines, it’s about unified DevOps platforms.

CloudBees Unify is a prime example: Launched in mid‑2025, it unifies governance across toolchains without forcing migration — an AI‑powered operating layer over existing tools. (Wikipedia) Other providers are doing similar strategic shifts, responding to DevOps teams tired of fragmented tooling.

Engineers say platform‑style tools work — when they’re flexible, extensible and don’t lock them out. Simplicity and integration win; fragmentation and vendor lock‑in frustrate.

Cloud Native: From Cutting‑Edge to Default

Cloud Native is no longer optional — it’s the de facto standard. According to surveys, 61% of organizations run some (41%) or most (19%) production workloads on Kubernetes, with AI/ML and analytics topping the list at 56% each. (TechRadar, devopsy.kinsta.cloud) Engineers affirm that masteries like ArgoCD, Flux, service meshes, eBPF and GitOps practices are now standard radiation of skills. (Reddit)

The link between Cloud Native and AI is especially tight; data workloads, model serving and feature rollout are deeply intertwined with container orchestration and microservices. DevOps engineers increasingly identify as “platform engineers” working at scale in that space.

Education and Certification: Catching Up Slowly

DevOps education and certification remain fragmented. Traditional certs — Kubernetes (CKA, CKAD), AWS/Azure/GCP and DevOps Foundation — remain staples. (Reddit) But DevOps engineers express frustration: Formal learning often lags behind real‑world tooling, AI integration, or platform engineering practices.

Many engineers now augment certs with hands‑on labs, bootcamps and informal community learning. Organizations are piloting internal platform engineer training programs to bridge skills gaps. Still, a mismatch persists between the modern tech stack and classroom syllabi.

Macro‑Economics: Jobs, Layoffs and Demand Shifts

In 2025’s turbulent economic climate, DevOps engineers feel mixed pressure. Layoffs in tech surged 35% in Q1‑Q2 2025 versus the same period in 2024. AI is partly blamed, especially in coding jobs.(Business Insider) But experienced practitioners remain in demand. Firms are automating low‑level roles while doubling down on strategic DevOps and platform roles. (Business Insider)

Waze co-founder Uri Levine recently argued that AI will actually increase demand for software engineers — especially those who can adapt, lead, and architect AI‑infused systems. (Business Insider) Meanwhile, others warn that entry‑level roles are shrinking—some startups now operate as “100× engineer” shops, using AI to super‑charge top performers. (Business Insider)

Most DevOps engineers I speak with say: keep growing skills, especially in AI orchestration, security‑by‑design, cloud cost optimization, platform engineering discipline. Senior and strategic roles are safe; apprentice roles are under pressure.

What DevOps Engineers Think of the Road Ahead

Looking forward, DevOps engineers express hope infused with realism:
- AI will accelerate productivity, but not replace engineers. They expect to supervise agents, shape workflows and solve complex problems—not type boilerplate. (Business Insider)
- Platform engineering is here to stay, but success depends on impact, metrics and a product mindset. Engineers expect more rigor in design, developer‑centric UX, service‑like governance and clear business alignment. (adaptavist.com)
- Cloud Native is standard, but emphasis is shifting to stability, cost control, observability and AI‑ready infrastructure. Knowledge of service mesh, eBPF and GitOps is now table stakes.
- Certs matter, but only to a point. Real‑world experience, continuous hands‑on exposure to AI‑augmented tooling and platform building are where skills live.
- Economic uncertainty looms, but so do opportunities — for architects, platform builders, security‑minded DevOps leaders and AI‑savvy engineers.
Final Thought

DevOps engineers today stand at a crossroads: Platform engineering and cloud tooling have matured into the ecosystem, AI is no longer experimentation but embedded flow. Job markets are shifting, but real demand remains strong — for creative, strategic and adaptable engineers who can shepherd tools, teams and AI together into scalable delivery platforms.

From my vantage, DevOps is no longer about “doing DevOps” — it’s about owning DevOps Platforms, empowering developers and leveraging AI as a force multiplier. That’s what the field’s elite expect, and that’s what you should be building toward.

Alan Shimel is Editor‑in‑Chief and Founder of TechStrong. He has been covering DevOps and digital transformation for over two decades, exploring the intersection of culture, tooling and emerging technologies.
July 28, 2025
Unifying VMs and Containers on a Single Platform

Why the Divide No Longer Makes Sense

When it comes to traditional vs. modern apps, the days of making a binary choice are long gone. Organizations must rely on both, as many traditional business-critical applications often run on virtual machines, while modern applications are typically container-based. Yet, inexplicably and to their detriment, many enterprises still treat virtual machines (VMs) and containerized apps as separate entities—splitting their infrastructure and platforms into two distinct halves. The problem? Each half requires its own licenses, its own management, and its own financial and operational overhead. This results in inefficiency multiplying at an alarming rate—slowing innovation, increasing costs, and creating operational headaches that could (and should) be avoided.

Imagine an IT world where traditional virtual machines and modern containerized apps coexist seamlessly on one unified platform. No duplicate infrastructure. No added costs. No separate teams managing different silos. No unnecessary complexity. It shouldn’t be hard to imagine. IT executives already recognize that future success demands a fundamental shift in how infrastructure is managed to drive agility, efficiency, and resilience–not just technology modernization.

The Advantages of a Single Platform for VMs and Containers

When you consolidate your workloads, you unlock a host of benefits. Below are three of the most impactful:

1. Consistent Operating Model

Managing VMs and containers on separate platforms results in duplication and inefficiency across the entire IT ecosystem:

●      Two sets of operations, policies, self-service portals, APIs, and skill sets.

●      Fragmented visibility, root cause analysis and troubleshooting across infrastructure and workloads.

●      Higher total cost of ownership (TCO) due to redundant systems and fragmentation.

In contrast, using a single platform like VMware Cloud Foundation (VCF) for both workloads delivers:

●      Unified operations, policies, APIs, and self-service portals

●      Holistic visibility across VMs and Kubernetes clusters

The result? A unified environment that allows more efficient app delivery and lower TCO.

2. Higher Asset Utilization & Better Capacity Planning

Fragmented infrastructure leads to resource underutilization and inflated costs. A unified platform improves asset use and planning by:

●      Run more VMs and Kubernetes workloads per server—requiring fewer physical servers to meet demand.

●      Reduce operating costs related to space, power, cooling, monitoring, and management.

●      Leverage efficient memory management and intelligent workload scheduling.

The impact? Increased server utilization, fewer hardware purchases, and simpler holistic capacity planning process.

3. Reduced Licensing Costs

It’s simple math. Running separate platforms for VMs and containers effectively doubles software licensing costs—forcing organizations to purchase, manage, and maintain two sets of platform licenses. A unified approach eliminates this redundancy by reducing licensing costs from two platforms to one—covering both VM and Kubernetes workloads under a single framework

The Time Is Now

Now is the time to converge your traditional VM-based and modern container-based applications, and VMware Cloud Foundation provides the opportunity to do so.

With VMware Cloud Foundation, unification isn’t just possible—it’s the only way to ensure long-term success. Customers have long trusted VMware’s superior virtualization capabilities to deliver on performance, availability, and security requirements for their mission critical VM-based applications. And with integrated compute, storage, networking, and management, VCF has long delivered everything enterprises need to modernize and optimize—without relying on bolt-on components or complex workarounds. The inclusion of vSphere Kubernetes Service, at no extra cost, provides a single platform to deploy, provision, and manage both virtual machines and container-based applications—at scale. No silos. No added complexity. No compromises.

Delaying this shift only adds to technical debt, increases complexity, and makes future migrations more difficult. But the organizations that embrace this shift today will be the ones leading the digital economy tomorrow.

Stop waiting. Unify your infrastructure, streamline your operations, and position your organization for sustained success and lower TCO. This isn’t a mere technical evolution—it’s a strategic decision.

April 10, 2025
Survey Sees Positive DevOps Outlook Driving Investments

A survey of over 500 DevOps practitioners finds that approximately 60% of respondents plan to boost investments in DevOps tools and platforms in the next two years, with 20% planning significant budget increases.

The survey was conducted by Techstrong Research, an arm of Techstrong Group, which is also the parent organization of devopsy.kinsta.cloud. Respondents indicated areas of investment that organizations are investing in most over the next two years are core DevOps tools and platforms, including testing, code artifact repositories, observability, automated deployment platforms, internal developer portals (IDPs) and infrastructure-as-code (IaC) tools.

Specifically, just under a third (32%) are planning to replace or upgrade their continuous integration/continuous delivery (CI/CD) platform in the next 12 to 18 months, while 23% are currently assessing their future requirements. However, only 16% are planning to transition to an integrated DevOps platform in the next 12 to 18 months, compared to 37% that already have. Additionally, only 20% have adopted platform engineering as a methodology for managing DevOps at scale.

Overall, the survey finds that 88% view the future of DevOps as generally positive, with half (50%) describing it as very positive. Only 1.5% of respondents had a negative outlook on DevOps.

Additionally, 62.5% said they view artificial intelligence (AI) as either an extremely or very valuable addition to DevOps workflows. Compared to 10.1% that view AI as either not being very or not at all valuable. Many organizations are already using AI in development today (32.7%), with another 41.5% considering using AI. Only 9% said AI is fully integrated into their DevOps workflows.

Mitch Ashley, principal analyst for Techstrong Research, said the survey makes it clear that while fads fade and other trends vanish, DevOps continues to deliver value to software teams and IT organizations. DevOps also has a long runway ahead regarding investments, with investments in automation, AI and application security driving a wave of additional investments, he added.

In fact, nearly three-quarters (72.7%) of respondents are applying DevOps on 50% or more of their projects, with 18.7% of respondents applying DevOps on all projects. The Build and Test phases of the software development lifecycle are areas that make the most use of DevOps practices at 78.6% and 72.4%, respectively.

A full 81% of respondents identified increases in velocity at which their organization delivers software into production environments as a top benefit of DevOps, with 45.6% describing those increases as being high. A similar percentage (80%) said DevOps also enables their organization to achieve faster times to market.

The survey finds a total of 81% also said DevOps improves developer productivity, with 35% describing those gains as high. DevOps teams are releasing code more frequently, with software delivered into production on a daily (11.8%), weekly (34.1%) and monthly (29%) basis respectively.

Despite those benefits, however, the survey also makes it clear additional improvements to DevOps workflows are needed, with survey respondent rating DevOps impact as low in managing complexity (16.3%), increasing collaboration between silos (14.8%) and improving software security (13%).

In general, the survey finds more than half of our respondents (55.8%) have achieved higher levels of DevOps maturity. Within this group, 34.4% are standardizing DevOps on projects across their organizations, while 21.5% said they have reached a mastery level of best DevOps practices.

In comparison, 18.1% indicated they are piloting DevOps on a subset of projects, while 21.8% are applying DevOps learnings across multiple projects.

The survey makes it clear that organizations continue to embrace best DevOps practices as they see fit. The one certain thing, however, is, to varying degrees, DevOps is now more indispensable than ever.

For more information, download a copy of the DevOps Next report here.

September 30, 2024
Engaging Developers in a Post-COVID-19 World

The time for platforms to rethink how they engage developers is now

How we communicate, socialize, look up information, work, shop and travel have all been transformed by platform companies. COVID-19 has only increased our reliance on these platforms, as people spend more time on the internet and businesses migrate from offline to online offerings.

These shifts present challenges to the platforms themselves. With people increasingly using low-latency applications such as teleconferencing and video streaming, there’s much more demand for data to be delivered without noticeable delays. As more and more varied businesses migrate their services online, it creates additional burdens on platform infrastructure and stretches the limits of their functionality.

It’s a challenging environment for platform companies. But disruption also presents opportunities. The question becomes, “What do platform companies need to do to remain relevant and to improve their products and services?“

The Importance of Developers

There are some obvious starting points. Platform companies will need to rapidly adapt their offerings according to their evolving user base. They’ll also need to plan for commercial sustainability and make sure that they balance short-term expenditure with longer-term profitability.

But they will not be able to do these things alone. With almost 24 million developers in the world today, external developers are central to any platform’s effort to adapt their products and expand commercial sustainability. As the technology industry continues to be subject to disruption, the developer ecosystem is ideally placed to drive expansion at a sustainable scale, enable product differentiation, reduce time-to-market and ensure relevance through product-driven growth.

Those platforms that can successfully engage developers will stand out from the crowd and win the increasingly competitive fight for developer loyalty. Similarly, as developer ecosystems increasingly become a key differentiator, the platforms that fail to engage developers may also fail to keep up with the rapidly changing technology landscape.

Developing Developer Partnerships

It’s clear that platforms will need to make engaging developers a priority moving forward. The question is, how?

For a start, organizations need to have a clear sense of which parts of their ecosystem can be opened up quickly and which integrations need to be built into the platform. With these established, the next step is to engage the developers that are best placed to help.

At the core of engaging developers successfully is the recognition that different developers will have different priorities, needs and core skillsets. All developers need to secure ROI by engaging with the platform and its ecosystem. But their perception of value will differ. Enterprise developers, for example, rate ease of use, peer perception, strong developer tools, accurate content and enterprise-level support. Independent software vendors (ISVs), however, care most about revenue, lead generation and being pulled into deals.

Segmentation, Segmentation, Segmentation

Investing in areas that will attract the right developers is key to long-term sustainability. And there are a few key areas that will require especially close attention. First and foremost, developer segmentation will be absolutely vital for every platform. As the product landscape expands, platforms need to be clear which new developers may want to engage with their platform.

By revisiting developer segmentation, new product use-cases can be included to ensure that content, support and messaging match what developers need. For example, if low-code/no-code developers are a new demographic, platforms need to extend content the scope of their help materials to include more basic explanations (including generic technology terms) and extra links to FAQs.

On the other end, platforms also will need to invest in tools that match the needs of different developer groups. Experienced, full-stack developers using the latest languages and patterns will expect sophisticated tooling, services and frameworks as well as DevOps tools.

Either way, continued focus on self-service will ensure cost-effectiveness by allowing platforms to manage extra usage without the need to apply extra resources.

Scaling New Heights With the Cloud

If transitioning critical services to the cloud isn’t already on the agenda for every platform business, it should be. A seamless service transition to cloud, with future-proofed expansion capabilities, is imperative for any platform that wants to engage developers. Why? Because as well as supporting service-based pricing, transitioning to the cloud facilitates scale and allows developers to access the platform securely from any location. Above anything else, a robust cloud infrastructure makes platforms more flexible, facilitates collaboration between platforms and developers and enhances the platform’s overall performance.

Thriving With Developers in a Post-COVID World

The time for platforms to rethink how they engage developers is now. Building a developer ecosystem will be vital for any platform wanting to scale at speed. If equipped with powerful tools, enabled through self-service and supported through effective educational resources, developers can be the difference between platforms that struggle in a post-COVID-19 world and those that thrive.

November 3, 2020
Building a Services and IoT Platform with DevOps, Part 1
With the growth of the internet of things (IoT), expected to connect more than 50 billion devices (things) in next three years, building intelligent systems in a more connected world is presenting many challenges. Especially in building software for IoT, beyond hardware and network connectivity challenges there are numerous challenges in development, test, release, managing the infrastructure, monitoring the infrastructure and devices (things) and security.

This article focuses on key components in IoT platform, how to select right set of DevOps tools for building those components, and creating a DevOps pipeline for the IoT platform.

Building Your Own Services and IoT Platform with DevOps

The main components in IoT solution comprises of device registration and identity, data ingestion, command and control, rules and actions for hot data analytics and cold data analytics, data storage, alerts and reporting.

Device Registration and Identity: The devices/edge need to connect to each other and to a cloud gateway to send/receive the data, for which the devices have to be uniquely identified and registered at the gateway so that only authenticated and authorized devices/users can communicate to the cloud. Every IoT platform vendors has its own way of device registration and identity methods for storing device metadata such as AWS Things Shadow and AWS Gateway Rules Engine, Device twin and Azure IOT hub in Azure, GE Predix, etc.

Data Ingestion: The devices communicate to the cloud/field gateway, which further ingests the data (device info and telemetry info, such as temperature data from sensors) into the cloud. You can use multiple options of PAAS services from Azure/AWS/GE predix, etc. for this component.

Command and Control: These components cause the device to take an action and send messages to a device from the cloud, which can be built with PaaS services such as AWS Kinesis, Lambda functions, Azure functions, Azure Logic Apps, etc. or open-source technologies.

Rules and Actions: These components act on specific device-to-cloud data, using the rules defined for processing hot data (where the data needs to be queried and analyzed within specified time interval) and cold data (where the data can be stored for delayed querying and further analysis). These components can also be built with existing PaaS services or using open-source tools such as Apache Storm, Map Reduce, Hive, Data Lake, etc. for data streaming and analytics.

Data Storage: You need to store multiple types of data ranging from device metadata, telemetry data from specified set of devices, hot data and cold raw data, which needs to be stored and scaled up based on demand.

Alerts/Notifications: The analyzed data needs to be visualized to derive more useful business metrics and outcomes, which can be integrated to external systems or other business applications to take preventive and predictive actions. Customized dashboards and data monitoring tools needs to be integrated to the solution to enable monitoring, alerts and notifications.

With increased complexity in IoT development, the large number of devices to be monitored and large volumes of data to be analyzed and to manage complex infrastructure, DevOps becomes essential to the IoT development life cycle.

Selecting the Right DevOps Tools for an IoT Platform

Selecting the right DevOps tools for an IoT platform can help overcome challenges in creating/pushing applications and firmware updates to large scale of devices, as well as monitoring and infrastructure provisioning on-demand.

With PaaS services used for IoT development, there is less code development but more configurations and rules-based development for developing key IoT components.

Also, the teams developing device software and application development may be distributed, but the releases need to be streamlined with complete traceability and auditability. You need to manage a scalable infrastructure to support an IoT platform.

Given below are some of the DevOps Tools used in most of IoT developments to help DevOps teams develop, build and release an IoT platform seamlessly.
- Version Control: Git, for distributed teams and maintaining repositories for device development, application development and infrastructure automation development. Other tools are Clear case, SVN or TFVC.
- Continuous Integration: Jenkins, VSTS (Visual Studio Team Services) and Bamboo to create jobs to do build, unit test and code quality analysis for the application, device software and infrastructure scripts, with the support of large collection of tools plugins and usage of different build tools such as gradle, maven, msbuild, etc., as well as supporting multiple technology stacks.
- Infrastructure Provisioning and Automation: Ansible, ARM (Azure Resource Manager) and AWS cloud formation templates to automate the infrastructure in multiple cloud vendors and use tools such as Terraform or Spinnaker to automate the provisioning of infrastructure across hybrid clouds and maintain a common infrastructure blueprint, which enables the porting/migration of infrastructure workloads across clouds.
- Infrastructure Testing: When PaaS services are used for developing certain components in an IoT platform, testing those PaaS components is a bit different—PaaS services have less code but more configurations and endpoints from dev/test team perspectives. Testing these components will involve endpoint testing, whether rules are created properly, how data flows between the different components and robust infrastructure verification. These infrastructure testing scripts can be written using Python/groovy or Vagrant to simulate the environment and test, or by using testing frameworks such as Pester for PowerShell scripts. You can extend your test solutions to leverage the REST APIs/CLI commands, exposed by different PaaS services, and these test scripts can test the PaaS services too, with simulated test data and generate the test result reports.
- Continuous Deployment: VSTS Release manager, Jenkins, Capistrano, CA Nolio, etc. can be used to create the deployment workflows for deploying device firmware updates, application updates separately with control to release on multiple environments, gated checks and streamlined deployment process, with feedback enabled.
- Test Automation: Selenium, Jmeter, security testing tools such as Veracode, EMMA, CA Lisa, Inspec, Robot framework and performance test tools are used for test automation. Testing in IiT is complex, as it needs an end-to-end testing of multiple devices interconnected. Most of the other device development may not be in your control, except your own device development, so to test an end-to-end functionality you need to have robust test cases and simulated test devices and test environments. This is where containers can help in test automation—a simulated device as a Docker container can run in one test environment and testers can use service virtualization tools such as IBM Rational Test Virtualization Server, CA Service Virtualization, Parasoft Virtualize or HP Service Virtualization to test their end-to end functionality when other dependent components/devices are being developed or not available.
- Security and Code Quality Analysis and Security Testing is key in IoT development. As more and more devices are interconnected, the risk of data breach/data privacy is high; hence, a shift left of security testing during the development phase as part of the CI pipeline is inevitable. Tools such as Veracode, EMMA and Inspec can be used to run security analysis and identify security vulnerabilities and compliance issues. Unit testing tools such as junit, xunit, nunit or mstest can be used based on the technology stack. Code quality analysis tools such as SonarQube, coverity, cobertura and Fxcop are widely used for code quality analysis in IoT application components development.
- Containerization and Container Orchestration: You can deploy your application as Docker containers, which have less footprint in the physical/hosted infrastructure and runs only with required application framework, libraries and dependencies. These containers/cluster of containers are managed with container orchestration tools such as Kubernetes, Docker Swarm, AWS Container Services, Azure Container Services, etc. As mentioned above, simulated test devices can run in docker containers, which enables shipping/moving the containers across any cloud and making the test environments ready within minutes.
- Monitoring: Monitoring in IoT involves monitoring the devices; infrastructure VMs/clusters; telemetry data; the build, test and releases; and billing, among others. Metrics such as cluster usage, CPU usage, network availability, response time, number of devices connected and devices updated/not updated can be configured and viewed in monitoring tools including AWS Cloudwatch, Azure Application Insights & LogAnalytics, Stream Analytics, New Relic, Prometheus (open source), Nagios, Coscale, SysDig, Dynatrace and GE Predix Monitoring. The tools can be selected depending on the target platform on different clouds or on-premises, technology stack and other factors.
- Cloud and Cloud Management: Cloud providers including AWS, GE Predix, Azure and GCP provide out-of-the-box features and support for IoT platforms with starter kits, IoT device SDKs, protocols, gateways customization and more. Cloud management tools such as Right Scale cloud management platform, OneOps, etc., can be used for managing the infrastructure workloads; bringing in central governance, access control and security; monitoring billing and optimization; managing security and standards; and streamlining and orchestration of IoT platform development.
Depending on the technology stack, the DevOps requirements and business needs, the right tool needs to be selected to streamline IoT platform development with feedback flowing between the different stream of teams involved in IoT development, and to bring in standardization in security, governance and tools, metrics to measure and to better manage IoT platform.

Release, Rollout Firmware/Application Updates

Rollout of firmware updates for an IoT platform is more challenging compared to application releases; with device updates, we have to deal with network latency, issues with connectivity of devices, security issues, different sets of devices, different deployment channels, deployments done over the air and different sets of protocols used by different manufacturers.

There is additional complexity involved when there are new additions to devices or when supporting legacy devices, and this needs incorporating integrated end-to-end testing as part of deployment pipeline, to test the devices and applications from a user-centric approach.

In IoT development, there are multiple development and deployment pipelines for firmware updates, server-side code, mobile apps and multiple devices delivering to multiple endpoints. Hence, the release cadence of IoT components varies, as application code updates can be released monthly, whereas firmware updates to devices can happen every three to six months—hence, a well-defined process must be followed for regularly integrating the code.

Continuous testing involves infrastructure testing, performance testing, end user testing and functional testing, all of which must be integrated with the release pipeline, where deployments can happen to simulated test environments, provisioned as environments as code (environment with all runtime dependencies needed to run the application like Docker containers), rather than simply infrastructure as code.

DevOps teams need to keep track of releases and monitor all these deployments, ensure environments are stable, trace back the releases with the device firmware updates and track the defects. The DevOps teams should pull the release packages of different components from multiple pipelines and prepare main and customized builds to integrate and deploy the components together.

The following are the main steps in CI/CD pipeline for IoT platform development.
1. Code checked into version control: Check in the code for application development, device/firmware development, PaaS services configurations, infra automation scripts and templates to the version control, pointing to the defined repositories.
2. Continuous Integration: Build jobs created in a CI server such as Jenkins/VSTS/Bamboo executes build compilation, unit testing, code quality analysis, automated reviews, environment configuration preparation, running security testing analysis, creating deployment packages or baked-in container images, for deployment.
3. Triggers Deployment: The build agents generate the packages for firmware updates application updates and notifies release agent to trigger deployment.
4. The deployment workflow can call a service/worker job, which in turn can connect to the cloud gateway services, to notify firmware updates.
5. The cloud gateway picks up the firmware updates and notifies devices for devices updates. The devices will the pull the updates and installs the updates and further send telemetry data.
6. Multiple testing environments in cloud can be integrated to the release pipeline to simulate the firmware updates.
7. In case of application rollouts, there will be canary releases to gradually roll out to different environments in different regions after functional testing/regression testing.
8. Testing: Security testing, firmware updates targeting to specific devices, performance of the device during the updates, device identity and memory usage, etc., needs to be addressed during deployments in IoT. Once the application is deployed to the test environment, functional/regression testing occurs and generates the test reports.
9. Monitoring metrics such as memory usage, performance of the device, network traffic, latency, devices connected, reliability, etc., needs to be monitored, as these are of prime importance to decide on how to implement remote updates and monitoring without affecting the performance of the devices and these metrics has to be notified to the Dev/QA/DevOps teams as part of release pipeline execution.
10. On-demand environment provisioning scripts/templates must be integrated with deployment pipeline to provision the test environments, add more devices as per the scaling metrics.
Establishing a CI CD pipeline well ahead at the start of development and following mature DevOps practices such as security testing, infrastructure testing, containerization, monitoring, cloud and DevOps governance and feedback-driven development are the keys to help address the challenges in building your own services and IoT solutions.

About the Author / Lavanya Subbarayalu

Lavanya Subbarayalu is Senior Architect working with Technology Office in HCL Technologies. She has expertise in IOT, Azure, DevOps consulting & Microsoft technologies. She is associated with DevOps COE, working on design and Development of DevOps solutions and consulting tools. Connect with her on LinkedIn.
October 25, 2017