While a lot is being written about cloud computing, it seems much of the coverage amounts to forecasting, star gazing, and outright dreaming. I’d argue that what’s needed now is a healthy dose of pragmatism.
The benefits of cloud computing are pretty obvious to even the casual industry observer: flexible scalability, cost savings, increased business agility, and the list goes on. While a lot is being made of this potential -- and a lot of media space gets dedicated to the future of this computing paradigm -- precious little is being written about some of the practical, near-term issues that confront those organizations considering a move to cloud computing.
In this article, I’ll take a look at some of those issues. First, however, some context is in order. While definitions of cloud computing continue to morph and be debated, there are three high level approaches to cloud computing:
- Application Platform Clouds. These services can include social networking applications, on-demand service desks, and CRM systems. While these offerings do meet the definition of cloud computing, most of these services emerged under another name, namely “software as a service” (SaaS).
- Private Clouds. This is really a description of a highly virtualized, on-premise data center that is constructed as if it were that of a public cloud provider.
- Infrastructure Clouds. This refers to pay-as-you-go computing infrastructure components, such as servers, data storage, and network entities. These services are delivered in billable units of consumption, in a manner very analogous to such traditional public utilities as electricity or water providers.
Infrastructure clouds have the potential to usher in the most disruption in the industry, so here I’ll be focusing on this approach. For IT organizations considering a move to infrastructure cloud computing, following are five key things to consider, before any move is made.
- The Cloud is a Third Party Service -- SLAs and Clear Accountability will be Vital
While cloud computing is a new, rapidly evolving paradigm, in the end it is fundamentally another third party service, and must be managed as such. Toward that end, it will be important to establish concrete, measurable service level agreements (SLAs). While these set an effective groundwork, the degree of reliance an enterprise places on a cloud provider can’t be overstated: if a vendor’s cloud goes down for two hours, some organizations could stand to lose millions -- lost revenue that SLAs simply won’t cover. Yet without clear SLAs and an effective means for verifying compliance, the damage from poor service levels will only be exacerbated.
- The Performance and Availability Will Be out of Your Control -- Monitoring Will be Key
The assets and infrastructure that support the cloud service are not owned by your organization. They are owned by the cloud providers -- organizations more likely have a focus on maximizing their revenue than on optimizing the performance and availability of your services. Outsourcing delivery of service does not imply that you can outsource the monitoring of that service. This has been true of any outsourced service provision, from the traditional managed service provider model to the application platform cloud model. Besides, today very few cloud providers are offering any form of service level monitoring to their customers. Establishing sound monitoring practices for cloud services from the outset will yield significant benefits in the long term.
- The Cloud is Relatively New -- Try Before You Buy
Even if services are being offered by older, more well-established vendors, the reality is that cloud service offerings are still relatively new, with very little track record. Often, service providers have a history of providing services that bear very little resemblance to the practice of service management. It is true that most leading cloud service providers have a history of providing some form of highly utilized consumer service based on massive computing facilities, but that is no reason to assume that they will provide the service that your business will need, either immediately or in the years to come. Consequently, piloting the services from multiple vendors is likely to be the best starting point.
- Security of the Network is not Guaranteed -- Assume the Worst
The use of cloud computing brings the security of interconnecting networks in to considerable doubt. Moving from a highly secure data center model to an Internet-based cloud model will require a complete review of network security. In essence, you have to assume that the network is totally insecure and that the only security mechanisms in place are those that come prepackaged with the infrastructure components in use. This has implications for all of the applications that you host on the cloud platform and may require a complete review or even a rewrite of those applications.
- Availability of Cloud Components is Irrelevant -- It’s All About Service Levels
This is the most significant systems management paradigm shift of them all. Most people’s systems management experience has lead them to associate systems availability with service availability; that is, if a network component was running normally, the services running across that network component were also running normally. There has been an implicit association of these two in most traditional performance and availability monitoring solutions. However, the highly dynamic nature of the computing infrastructure that exists in the cloud makes traditional CMDB (or simple list) based systems management virtually impossible to implement. All the traditional server and network reporting that shows 99.999 availability will become secondary and probably irrelevant for future service level management and reporting. What this means is that synthetic transaction monitoring -- that is, generating, monitoring, and reporting on simulated service requests -- will be of paramount importance.
Make no mistake: cloud computing can, and in fact already has, delivered some real, bottom-line benefits to organizations of all sorts. However, it’s also vital to understand that, contrary to a lot of the hype, these benefits are anything but assured. Cloud computing represents a fundamental paradigm shift for providers and customers alike. By taking some of the above factors into consideration, an organization can be better poised to mitigate some of the risk, and more fully capitalize on some of the benefits.
Mark Rivington is vice president of technology at Nimsoft.