Three years since the blockbuster acquisition, LinkedIn is still running its own datacenter infrastructure. But that is all set to change, as migration to the Azure cloud platform will begin soon.
Many other Microsoft sites and services have adopted Azure.
But even though LinkedIn evaluated moving to the public cloud many times over the years, the team felt like it was not the right time for making the jump. As detailed, it was only a couple of weeks ago that the team finally pulled the trigger on what sure seems like a significant migration effort.
One that will be spread over many a year.
The business-centered social network has 645 million members, up almost 50% from 2016 when it was purchased in an almost $27 billion deal. The Azure public cloud is now at a level when it can be entrusted with the critical data of its members.
As Mohak Shroff, senior vice president of engineering at LinkedIn, notes, the shift will take place over multiple years:
“The cloud holds the future for us and we are confident that Azure is the right platform to build on for years to come.”
Everything on the LinkedIn infrastructure is now moving to Azure with this effort, though the primary focus is to move members over slowly, so as not to compromise on the accessibility, reliability and performance of the website.
The move is expected to be cost-neutral to LinkedIn, and company officials believe that it will produce efficiencies over time.
Inevitability, you say?
Azure was the obvious choice, but that does not mean LinkedIn did not take a look at other public clouds. The company considered them in the past, though in this instance it did not. The primary reason being the fact that LinkedIn is a fairly complex case for the public cloud, with specific custom needs.
That said, there are notable benefits for the move.
First being the site becoming more proximate to its members thanks to the number of Azure regions. Second, the elastic nature of the public cloud, particularly at the needs for bursts. Thirdly, the ability to focus on building products on the Azure stack over time, and build customer facing functionality.
Sounds like a plan.
What do you guys think?