In the high-stakes world of cloud computing, even big tech giants don’t always get it right. LinkedIn, the professional social networking site owned by Microsoft, recently made waves by quietly shelving a years-long effort to migrate its infrastructure to the Azure cloud platform. Codenamed “Blueshift,” the project represented a major bet on Microsoft’s cloud capabilities but faced technical and resource challenges that ultimately forced LinkedIn to change course — at least for now.

In 2019, LinkedIn engineering VP Mohak Shroff boldly announced the company would undertake its most ambitious tech migration by moving entirely to Azure, which Microsoft execs hoped would showcase Azure’s power and scalability. Internally dubbed Blueshift, Shroff said the standard Azure tooling would let LinkedIn “innovate faster” and unlock “unprecedented scale.” But three years later, the tune has changed. Documents and insider accounts suggest that instead of utilizing Azure’s turnkey services, LinkedIn insisted on using its custom tools and architectures — a decision that proved difficult to execute at Azure’s expansive scale. Rather than setting the standard for Azure migrations, Blueshift became a cautionary tale about the perils of overcustomization.

In June 2022, a somber memo from LinkedIn CTO Raghu Hiremagalur acknowledged to engineers that the company would “pause our planned migration to Azure” and instead “focus our efforts on scaling and innovating our on-prem infrastructure.” Insiders say Azure’s skyrocketing growth made dedicating resources to Blueshift untenable, prompting Microsoft and LinkedIn to agree to postpone the ambitious initiative jointly. LinkedIn seems content keeping most workloads in its data centers while utilizing Azure for supplementary needs. Far from the full-scale lift-and-shift migration Blueshift envisioned, but a pragmatic response to technical realities.

With the dominant AWS facing an increasingly fierce challenge from fast-growing Azure, Microsoft was surely hoping to showcase LinkedIn as a marquee Azure migration story. Its reversal serves as a warning that even Microsoft properties won’t always bet the farm on its cloud platform. But Microsoft’s cloud growth remains astronomic, thanks to booming demand from AI models like ChatGPT. If Microsoft stays focused on that explosive growth, perhaps deprioritizing resource-intensive migrations is wise — even if it means Azure won’t be powering the likes of LinkedIn just yet. For LinkedIn, de-risking with a stepped migration approach may disappoint Azure’s boosters, but keeps their infrastructure plans firmly grounded. Because in the cloud game, it’s better to travel hopefully than to arrive outright crashed.