One foot in the cloud, another on the ground
- Summary:
- Leading SaaS vendors want you to move everything to the cloud, but most of them keep firm control over their own core infrastructure. Enterprises can learn valuable lessons.
The explanation for this apparent paradox yields valuable insights for any enterprise considering whether to move computing to the cloud or retain it in-house.
If what the cloud vendors do (as opposed to what they say) is any guide, then here are the criteria that swing the argument in favor of keeping computing in-house rather than using third-party infrastructure:
- The processes involved are unique to you and fundamental to your offering;
- You operate at sufficiently large scale to realize economies or performance benefits by running your own infrastructure;
- Your customers have confidence in your ability to provide a reliable, durable and competitive service using your own resources.
This is not to say that cloud vendors don't use third-party cloud applications for aspects of their operations that are ancillary to their core software offerings. In fact, their use of each others' products for the likes of CRM, HR, travel and expense management, collaboration and other common business processes is so pervasive that it's almost incestuous.
What cloud vendors do
But when it comes to their own software offerings that form the core of their business, here are some specific examples of how well-known SaaS and cloud vendors implement the above guidelines:
- Google and Facebook not only operate their own infrastructure, they go as far as building and operating their own datacenters, which they populate with servers built to order to their own custom designs. There couldn't be a stronger confirmation of the message that, if you have a big enough requirement, then nothing beats your own infrastructure.
- Salesforce.com operates its own custom software stack that is highly tuned to deliver maximum performance to all its customers. I could have cited any other large SaaS vendor as a similar example — Workday, NetSuite, Concur and many others all operate their own product stacks. They do so because they have been able to build an application architecture that maximizes the pooling of resource usage across all their customers to such an extent that it would never be cost-effective to use a thid-party platform. It's especially interesting to look at Salesforce for two further reasons:
- Other vendors run on its platform — we'll discuss this below
- Its Heroku offering is an exception to the rule, hosted on Amazon Web Services. This remains economical because the service Heroku offers has more potential for sudden peaks and troughs in usage and therefore the economic justification for Amazon's pay-as-you-go model is stronger.
- FinancialForce.com is one example of the many vendors that opt to run on the Salesforce.com platform rather than their own stack. This is typical of vendors with smaller vendors or those still in start-up mode where the costs of building and operating their own infrastructure would be prohibitive. In addition, customers often feel reassured to know that the infrastructure of a younger, growing company is operated by a larger, more established player.
- Sage is an example of an established software vendor that has chosen to use third-party cloud infrastructure for its SaaS offerings. This allows it to scale up from a lower user base without the large upfront costs of investing significantly in its own cloud skills and infrastructure. Many SaaS vendors start out on third-party infrastructure but opt to go onto their own infrastructure later on once they have validated the market or reached an operating scale where it is more economical to run their own infrastructure.
Hybrid infrastructure
The takeaway from all this for enterprises outside the software industry is that even pureplay cloud vendors operate a hybrid infrastructure. They use third-party cloud applications for ancillary aspects of their operations, but as soon as they have sufficient scale and volume of usage, they choose to run their unique core processes on their own infrastructure.
For many businesses in the real world, of course, their core operations rest in manufacturing, retail and distribution, hospitality or some other activity that relies on physical infrastructure rather than computing.
For these companies, it may still make sense to rely on cloud providers for their computing, because this does not affect their core business infrastructure. This indeed is the business opportunity targeted by Fujitsu's cloud platform strategy. The cloud platform aims to add more automation and cloud monetization options to those core processes.
But for enterprises that rely significantly on software to operate their core processes — and if Marc Andreessen is right about software eating the world, then there will be more and more of them — the example provided by pureplay SaaS vendors suggests that in-house computing assets will remain a reality for these organizations many years into the future.
Disclosure: Salesforce.com FinancialForce.com and Workday are diginomica premium partners.
Image credit: © silent_47 - Fotolia.com