Tech Debt and Migration Management

3 posts · Nov 2019 – May 2020

What this trail covers

Spotify's approach to managing technology debt and migrations has evolved to minimize engineering team exposure and reduce fragmentation. This involves a company-wide migrations map, productizing migrations with dedicated PMs and automation, and extensive automation to abstract underlying infrastructure changes. The goal is to enable faster innovation by moving up the stack and leveraging cloud capabilities.

All Spotify trails

Jun 2020Tech Migrations, the Spotify WayThis post details Spotify's strategy for tackling technology migrations and reducing tech debt. Key contributions include: establishing a company-wide migrations map with negotiated priorities and deadlines, consolidating communication through Backstage, adopting a product management approach for migrations with dedicated PMs, alpha/beta testing, and internal training, and leveraging automation to abstract infrastructure changes and send automated PRs. The 'Tech Insights' plugin in Backstage is highlighted for gamifying migration progress.
Apr 2020When Should I Write an Architecture Decision RecordThis post introduces and advocates for the use of Architecture Decision Records (ADRs) as a lightweight documentation practice to capture significant technical decisions. It outlines the benefits of ADRs for onboarding, ownership handover, and team alignment, and provides a decision-making framework for when to write an ADR, covering backfilling undocumented decisions, proposing large changes (often following RFCs), and documenting small but impactful changes. The post emphasizes that ADRs are a crucial tool for managing technical debt and ensuring consistency in system design and engineering best practices.
Dec 2019Views From The Cloud: A History of Spotify’s Journey to the Cloud, Part 1This post details the first part of Spotify's journey to the cloud, focusing on the strategic decision-making and initial migration phases. It covers the rationale for moving from on-premise data centers to Google Cloud Platform (GCP), the evaluation of different cloud strategies (multi-cloud, hybrid, single provider), and the choice of an 'all-in' approach with GCP. The post describes the pragmatic mix of 'lift and shift' and rewrite strategies for user-facing services and data processing, respectively. It also highlights the management approach, including a migration visualization, standardized sprint program, and teaching materials, leading to the successful routing of traffic to GCP by May 2017 and the retirement of on-premise data centers by 2018.
Advertisement