Are enterprises flocking the public cloud?

Now, this might sound cliché! But everybody is running to the cloud! Going by recent quarterly reports by Amazon, Microsoft and Google, their respective cloud revenues have grown at a tremendous rate—combined annualized revenue for these three public cloud juggernauts was more than $45 billion based Q417 reported revenues. According IDC, public cloud revenue is expected to reach $160 billion in 2018 and expected to keep up the same growth for the next few years. Bulk of this is coming from SaaS which is nearly two thirds of all public cloud spend. The SaaS phenomenon is not new. Every enterprise uses multiple SaaS clouds for HR, ERM, CRM and IT operations management. in short, enterprises are flocking to SaaS.
Same is true for PaaS and IaaS! According to IDC, PaaS is the fastest growing segment with 50%[1] growth coming from application development and deployment automation toolset for cloud native applications. On the other hand, IaaS growing at 38.1% representing 17% of the total public cloud spend. It is clear, enterprises are flocking to PaaS & IaaS as well!

What’s driving this migration? Which applications, what kinds of use cases? Are the customers running existing enterprise application in the cloud? Did they have to change the applications to move to the cloud?

The majority of companies running their business on the public cloud are small companies—and most of them are new SaaS companies. Large enterprises are using the public cloud for non-core apps and some specific uses cases like DevOps, big data workloads, analytics and so on. Compute intensive workload requiring elastic compute are great candidates to exploit on-demand “pay as you go” model and elasticity of the cloud. This would explain where the public cloud PaaS/IaaS revenue is coming from.

But how about core enterprises applications? It is an undeniable fact that public clouds have opened up new avenues for architecting IT using the capabilities that were hard to access (e.g., ML/AI, DR, cloud compute elasticity).  Are the enterprises in a position to exploit all of these capabilities? While some workloads are easy to move to the public clouds, others may have some barriers related incompatibility of public clouds, data governance, sovereignty, and security & recent private cloud investment. For example, due to sovereignty rules, certain kind of data cannot be stored outside private data center or geographical location. In many cases, private cloud may be the best option even economically like in the case of Dropbox, which saved $74.6 million over 2 years, by creating more optimized infrastructure for its service.

Public clouds come in different sizes and shapes. Not every cloud is a good choice for all the applications. For example, while AWS has best coverage and scale for new age apps, GCP offers the best ML/AI capabilities. Azure is most suited location for Microsoft applications and IBM offers enterprise-grade. Customers may also experience choice limitations due to location restrictions, & regional compliance rules. For example, Canadian insurance institutions have to use public cloud based in Canada. Likewise, regional players like Alibaba could be the only option for many companies in China for certain applications.  Then there are other issues like vendor lock-in, application performance, changes required for applications, data management across private and public clouds and so on.

To deliver applications and services and remain competitive in the marketplace, enterprises may need to use cloud services. However, their decision depends upon a set of questions.

  1. Which cloud model is better—private or public?
  2. Which public cloud is the better choice?
  3. Do I need to change my existing business applications to move to the cloud?
  4. Would I be locked into one cloud?
  5. Do I need the capabilities offered by different cloud vendors?
  6. Does public cloud make economic sense?

Enterprises now need to make decisions based on all these factors & select best of breed cloud(s) for their use cases and applications. This by itself becomes a big data problem; to forecast benefits/gain/ROIs not only for today but also for the future considering ever changing public cloud economics and offered capabilities, data sovereignty rules, the need for potential migrations from cloud to cloud for leveraging newly offered capabilities and so on. Because of this, enterprises have not moved majority of their enterprise-specific business applications to the public cloud.

It seems like most of the enterprise may need to embrace multiple clouds!

[1] Year over year in 1H17