The future of information technology migration strategy is about flexible SaaS software and IT infrastructure tools.
It’s amazing to see how much Information Technology has changed over the years. Back in the day, the first step in deploying new IT infrastructure or a new service would be to call IBM to order a physical server. You’d have to plan far enough ahead to allow for delivery and installation in your datacenter. You’d install the OS and all of the supporting libraries. Finally, you’d manually install and configure your code. You’d lovingly care for each of your machines as if they were your pets.
Today, we are rapidly moving towards the other extreme: either containers when you need servers or, ideally, SaaS that eliminates them entirely. Need a database? No problem, just use the API provided by your cloud provider—no server necessary. And with that, you eliminate not just the startup cost, but you also remove the often-overlooked ongoing operational costs (Deploying patches and OS updates on a critical, non-stop service is something I consider both fraught with risk and particularly odious).
In addition to outsourcing the necessary drudgery of system maintenance and feature enhancement, modern IT services are designed for automation. For example, consider something we did this morning: a service update, including simultaneous instances and live traffic migration. The entirety of human involvement in that process consisted of clicking the “start” button (In our development environment, we don’t even do that: this process is kicked off at check-in to master).
We, of course, have the advantage of being a startup. We began with a green field, and were able to jump right to the cutting-edge. My friends from large companies remind me that they have a large base of legacy technology and software that is serving live customers. This doesn’t prevent them from moving forward, though: the spectrum of benefits spans a wide range of hard and soft returns that far outweighing the cost of migration.
Migration isn’t usually easy, though. Moving foundational IT isn’t just a technology problem: it cuts across the organization and requires change management from many stakeholders. At this scale, you’re not just rebuilding the plane while flying it, you’re building an entirely new plane at the same time, then moving all the passengers across mid-air.
An example from one of our customers shows the complexity. Our customer is working on rolling out unified logging. They have a range of independent products that each use similar, but not identical, logging and telemetry stacks. Moving to a centralized layer allows for greater efficiency in data processing, as well as the ability to create a single, consistent reporting interface for business intelligence.
The justification is compelling, and the concept is straightforward. However, the implementation spans various product groups, each with their own release schedule, supporting a variety of different clients. IT needs to collect metrics from different operating systems running in datacenters that are distributed globally. Both Support and the CEO rely on the existing system, and they regularly issue change requests to collect new data and run new reports.
Many of us have had the experience of a “big bang” migration, and it’s likely not something we want to repeat. Choosing a phased approach reduces the risk, but it means coordinating activities across the organization. Changes need to be made to account for distinct project schedules. Stakeholders, with different strategies, metrics and success measures, must be kept updated, both to inform and to coordinate activities.
We’re helping our customer tackle this complexity with a custom “board” that includes both strategic and tactical dashboard elements. This is layered on a Kanban model that shows everyone the stages of the project, oriented on the different implementation themes. Each major task is represented as a rollup of subtasks, along with action items and prerequisites. The board is organized in columns that correspond to the current state so the big picture and details are always available: queued, preparing, in process, and done.
At a glance, everyone can see the current state of the migration, how they’re doing against metrics, and where they are in the queue. Critical milestones across all of the silos are connected to the process. This provides extensive visibility, not just into the state of the project, but into each task in context. Everyone is aligned.
How do you coordinate large scale projects across your own company? If you’re faced with a major IT migration project and need to define a process, we’re happy to share our experience and best practices. If you already have a process and need a tool that works across the organization, we can help with that too. We love to hear how people solve this widespread problem, so if you have something to share please reach out.