So it's time to choose an enterprise public cloud provider

This report looks at a few of the practical considerations in selecting and using a public cloud provider, something that is rapidly becoming a required exercise in present-day mainstream IT. Businesses and organizations that were born in the cloud or had the most to gain from moving early, like media and online social platforms, already know how to do this, of course. However, at this point IT enterprise-class, back-office or line-of-business applications, some of which have been central to enterprises for decades, are another story. Public cloud is a large and dynamic force in the IT market, growing faster than any other segment and measured in the tens of billions and there are undeniable benefits for a wide array of traditional enterprise applications as well as (and possibly more importantly) improvements to operations and business process.

The 451 Take

It is harder every day for enterprises to hesitate about using cloud services, which are, after all, all about making IT move faster and removing processes from the business that don't make any money. Most enterprises have already at least dipped a toe in at this point, but making it into a coherent strategy is still a lot of work. But the first considerations aren't technical – anyone can find out if their application will run in a public IaaS environment in a few hours and for a few dollars. It's just as easy to pilot a concept or a new project on public cloud. The larger, long-term challenges revolve around how the model of on-tap IT resource delivery fits into all the other moving parts of an IT organization. It's not as exciting as talking about Internet of Things (IoT) items or micro services, but for most enterprises it is far more necessary.

Cloud considerations

First, adopting and integrating cloud infrastructure services for enterprises often starts as a process of rationalization. Many enterprises want to pick a single provider and stick with it, just as they have done with traditional IT products, but 451 Research survey data shows that 66% of enterprises end up with more than one provider. It's important to consider this either as an existing reality or as part of an overall cloud transformation process. Consider the following:

  • Is single platform necessary or desirable and why?
  • Is there an appropriate policy and/or software-based way to manage multiple IaaS providers with the same set of operations guidelines and ensure consistent outcomes?
  • How is portability achieved or is lock-in acceptable as an outcome? If a significant amount of effort is required to make workloads ready for a public cloud provider, should that work entail not having to repeat it, or is that too difficult for beginning stages?

Second, for all the major public cloud providers – Amazon Web Services (AWS), Microsoft Azure, Google Compute and IBM Bluemix for this exercise – basic infrastructure readiness for these platforms is on par: compute, storage, availability, load balancing, virtual segregated networks (VPC), DNS service, and robust uptime and data replication. Objective, independent benchmarking is available for all of them, as is pricing. All discussion focuses on the practical implications to adoption relative to existing infrastructure, existing control systems and suitability for different workloads. It is also generally understood that enterprise stakeholders, especially development and application teams, will be familiar with these platforms and there will be a range of experience from novice to advanced, and pilots may have already been done.

Overall, the first questions to ask are the following:

  • How suitable is your current architecture (for any given workload) for use on a given public cloud and how much work is involved in migrating or adding public cloud resources to existing infrastructure and operations?
  • Network architecture is the corollary discussion. Is your current network suitable for accessing, managing and monitoring a connection to a public cloud, and if so, what requirements are there?
  • What is the status of current risk mitigation, business continuity and data compliance policies and what has to happen to ensure compliance or policy continuity when consuming public cloud resources?
  • How big a change to day-to-day operations will this entail after adoption and/or migration? For instance, how does this fit into existing control frameworks in use?
  • Can this be done in a sensible fashion? How much does this upend the normal way of doing things and can we do it without causing chaos?
  • What is needed in terms of non-IT resources like additional staffing or training?
  • Is there a need or a wish for expert third-party assistance like systems integrators or managed service providers?
  • And of course, how much testing has been done and is there a breakdown of the potential costs?

These are most of the important topics unrelated to technology that need to be addressed initially – individual cases will have additional, individual concerns, of course. The next consideration takes into account the much less objective topics of the aggregate benefits to each platform. Is there a significant foreseeable benefit to moving to any of these platforms because of their unique characteristics? These comparisons are for example only; the potential benefits or features of interest will vary for each organization.

AWS

The original platform that is more than the sum of its parts, AWS has a portfolio of services that is very broad indeed. Many of its services are stand-ins or replacements for standard enterprise software like virtual desktop, but there are unique AWS-specific services like Aurora, RedShift, and the MDM/IoT platform. AWS also has the ability to monitor, shape and massage its service platform with services like Cloud Formation and Lambda that may warrant long-term planning around.

Azure

Considering Azure requires a discussion of the benefits of Microsoft compatibility across platforms; for instance, does this entail AD integration at a company-wide level, and is that of substantial benefit to operations/administrators? Does that include Office 365 and other Microsoft-supported SaaS providers that Azure makes available via SSO? What does that do to the risk profile if end users and desktops, productivity apps and development resources are all under the same umbrella?

Google Compute

Google is already ubiquitous in enterprises; not only in search but also in mobile development and advertising, even if several steps removed from IT. Developers heavily invested in mobile or media development may find concurrency here with using Google's infrastructure services as well.

IBM Bluemix

451 Research survey data shows that IBM is rated as the cloud provider that understands enterprises the best, and Bluemix offers ready access to tools and services needed for enterprise computing, as well as proximity to Watson, IBM's machine-learning platform that is being integrated into a wide array of existing IBM products and services. 'Cultural compatibility' can also be a factor in choosing a cloud and shouldn't be overlooked.

Finally, if there is no impetus to pick one public provider over another, and all that's wanted is basic, on-demand compute and storage, it may be that the IT organization can find what it needs with a local managed infrastructure provider and eliminate much of the work of adoption and migration with professional assistance.

That is a summary of some of the important practical considerations in choosing a public cloud. It's complex and detailed, but it is necessary work and completely unrelated to technology. For many enterprises, the actual technology will consume far less time and effort in transitioning to cloud. 451 Research has several ways to assist with this process, including the Cloud Price Index (CPI), a comprehensive tool for estimating costs around IaaS usage, and 451 Research analyst-led advisory services.



Carl Brooks

Analyst, Service Providers

New Alert Set

"My Alert"

Failed to Set Alert

"My Alert"