There seems to be a trend of providers abandoning the commodity public cloud market. We saw HP exit its Helion Public Cloud, and more recently, Verizon shut down one of its Infrastructure as a Service (IaaS) products. At the same time, we see Amazon and Microsoft heavily committed to public cloud and making a lot of money. I think there is a fundamental difference between what the successful cloud providers and the commodity VM providers offer. The big difference is that successful cloud providers sell mostly non-commodity services. They sell services that are not available elsewhere. The value proposition for AWS and Azure is not really in running your VMs. It is in offering services that your applications, or users, can consume. These cloud services are consumed by application developers: information systems people rather than information technology. They lock in customers by delivering unique and valuable services. They have a low cost of entry to entice customers and a high cost of exit to retain customers.
A commodity cloud vendor will usually start with IaaS, providing VMs with an operating system. The value proposition here is that the provider can rack and stack virtualization hosts more cheaply than its customers. There are some additional values around consumption-based billing and OpEx vs. CapEx, but the big one is the economies of scale. This is really hard to make economically viable for customers with a reasonable number of VMs (20+) and VMs running production 24×7. The customer is still responsible for all VM operating system and application management. The rack and stack is a small fraction of the total cost of ownership of each VM, so potential total savings are not great. If your workload is not expanding and contracting a lot, then IaaS is likely to be more expensive than building your own virtualization platform.
One of the usual characteristics of the cloud services that have failed is requiring contracted billing rather than credit card billing. The customer commits up front to a regular monthly spend for several years. This high cost of entry inhibits adoption and makes IT the initial customer. The IT department must commit to using the cloud service rather than an individual developer or team choosing to use a service. Successful cloud services allow credit card billing with no contract. Getting onto the cloud is easy for any developer with a corporate credit card.
AWS is the king of cloud services. It started with storage as a service (S3) and message queueing as a service (SQS). Those were the first services: no VMs at the start. To use these services, customers needed to develop their own applications that talked to the services. Customers had to customize software to benefit from AWS. Over time, AWS added a lot of additional services: load balancing, databases, transcoding, and yes, even VMs. But the core value is that these services are used by customers’ custom-developed software. There is very little commodity here. You cannot lift an application that uses these AWS services and land it on another cloud. Microsoft has followed the same path, starting with SQL as a service and adding SharePoint. Its Office 365 product goes even further, as it is Software as a Service (SaaS). Using all these services is cheaper and easier for customers compared to building these services inside VMs. But using these services locks customers to a single cloud provider. The low per-month, per-service billing makes it easy for customers to get started. It just takes one developer with a credit card and an expense account to become an AWS customer. Sometimes IT only gets involved when the service goes into production and the bills start getting bigger.
There might be small customers who are satisfied with IaaS alone: those customers who require half a dozen VMs. It is very expensive to build a highly redundant virtualization platform for such a small estate. On the other hand, these same businesses are likely to simply use SaaS offerings. Microsoft Office 365 offers an Exchange email service without the need to maintain or patch Exchange Servers. Alternatively, Google Gmail and Google Drive between them offer another option for office productivity SaaS. These customers are then left with the need for a line-of-business application. These are increasingly available as SaaS offerings.
Public cloud is not primarily about commodity VMs. It is not about offering an open set of services that can be used on many different platforms. Public cloud is about applications that deliver great value to customers and lock those customers to one cloud provider.
Share this Article:
Latest posts by Alastair Cooke (see all)
- When Every Company Is a Software Company, No Company Is a Software Company - March 20, 2017
- Microsoft’s New Golden Goose Is Azure - March 16, 2017
- It Is in the Cloud—Who Cares If It Goes Down? - March 10, 2017