Vendor locking in the cloud

Getty Images

The "bring your own licenses" lift-and-shift cloud deployment model is giving way to more restrictions as software vendors look to steer existing user bases to their own cloud offerings.

As the government and industry move aggressively toward cloud computing, new software licensing restrictions are forcing customers to make hard choices between the infrastructure they want and the software they have. 

In this hyper-competitive environment, one tool to thwart competition is to leverage one's existing software install base by erecting licensing barriers that penalize customers for using competitive cloud offerings. 

This is what Microsoft appears to be doing according to multiple press reports and complaints filed with regulators. In October 2019, Microsoft changed its standard license agreements on its market dominant products, including Windows Server and SQL Server. The effect was to make it significantly more expensive to use those products in many cloud environments other than Microsoft's own Azure cloud. 

Prior to 2019, it was possible, with some accommodation, to simply lift on-premise Microsoft applications and shift them to any cloud provider without penalty. This was commonly known as the "bring your own licenses" lift-and-shift deployment model. You could run your already licensed Microsoft software in Amazon Web Services, Google, Oracle or any other cloud without concern. Not anymore.

Now, Microsoft explicitly forbids newer licensees from moving their already purchased on-premise Windows Server and SQL Server licenses to AWS, Google and Alibaba—some of their key competitors in the cloud space. To use those competitive cloud providers, Microsoft now forces consumers to repurchase new licenses. As Microsoft owns about 70% of the entire market for operating systems, the new policy gives Microsoft a unique opportunity to steer customers into its own Azure cloud, where new licenses don't need to be repurchased. 

"Microsoft is taking its arsenal of Windows Server, a massive software installed base, and using it punitively against its competitors," Gartner's Raj Bala told The Economist in 2020. The result has been a large spike in Azure's market share, suggesting something other than organic growth.

In a report in Bloomberg this week, Microsoft customers and resellers complained anonymously that the software giant was leveraging its huge install base to make it expensive to bring Office software into rival clouds. 

"There definitely are some valid concerns," Microsoft President Brad Smith told Bloomberg. "It's very important for us to learn more and then make some changes." 

The Bloomberg report isn't the first time customers have flagged the way licensing restrictions can distort cloud competition. In March, the Wall Street Journal reported that France's OVHcloud lodged a complaint against Microsoft alleging anti-competitive cloud practices with European regulators. 

There is history here. In the 1990s Microsoft was sued for similar behavior over its efforts to grow its Internet Explorer browser. Netscape had the market leading browser before Microsoft leveraged the desktop dominance of Windows by giving away Internet Explorer with Windows for free. Eventually this killed Netscape. The facts of the Netscape litigation track to a  similar pattern in respect to Microsoft's cloud aspirations. Smith represents a new Microsoft regime. Let's hope some lessons of the Netscape litigation have been learned. 

To be fair, Microsoft is not the only vendor looking to leverage a legacy advantage in the cloud. Other companies with a large installed base of on-premise software may be engaged in similar restrictive activities. For instance, one study showed that Oracle licenses its software in its own cloud on an "actual" CPU basis, while in a competitors' cloud, it licenses on an "available" CPU basis. The former licensing model may significantly reduce cost, providing an obvious advantage to use the incumbent's cloud. The temptation to engage in these types of restrictions, to thwart competition, may be hard to resist, particularly for software companies with huge, existing installed bases.

Why should the federal government care?

The roughly 2,000 pages of the Federal Acquisition Regulations are largely written to ensure competitive fairness and value for the government. The negative effect of decreased competition drives innovation out of the market and once competition subsides, the victors can extract higher profits—in this case, on the backs of American taxpayers. Furthermore, the lack of mobility in software starts a stealthy march toward vendor lock. 

To block the licensing provisions that thwart competition and create unfair advantages for incumbents, the government needs to attack this problem at the policy level. Although Congress mandated the creation of an "IP Cadre" in Section 838 of the National Defense Authorization Act of 2020, that focus appears to be on government agencies maintaining proper ownership of the software it develops or licenses to ensure continuity of operations. That's a related issue, but it's not the same as a close look at larger anti-competitive structural issues such as vendors using licensing to block competition. 

Congress could put a focus on provisions that limit competition and slowly destroy innovation. Ideally, the FAR and CIO Councils would work with Congress, the FCC, the new IP Cadre, the Department of Justice or appropriate other regulators to expose the cost of these practices to the government—and by extension all consumers—to promote a more competitive 21st-century software marketplace.

The time is now for the government to develop a strategy for license and cloud mobility to increase competition and promote innovation before it discovers it's locked into antiquated technology with prohibitive or even insurmountable switching costs.

Michael Garland is the founder of Garland, LLC, a consulting firm that advises clients on issues related to federal procurement law and the business of IT. His current clients include Alphabet, Grant Thornton and Octo.

NEXT STORY: Scene Graph Technology: High-Impact Use Cases for Government