Some argue that private Cloud computing is a natural evolution for IT modernization investments that enterprises continue to make over the years.
The risk with investing in private Cloud infrastructure, processes and organizational re-structuring is it is difficult to change course once the commitment has been made. Depending on the scale involved, the financial justification will be realized over a prolonged time period and during this time there may have been some innovations in Cloud architectures that leapfrog how private Clouds are architected today. There is a high level of uncertainty around this given the rapid snow-balling effect of Oracle, HP, Dell, Cisco and IBM all continually innovating to provide ‘Cloud in a Box’ solutions and we anticipate the market getting more and more competitive as Moore’s law continues to take effect.
In many cases, the best solution is to then go for a hybrid model, one in which you still gain the benefits of having a dedicated Cloud environment, but leave most of the ‘heavy lifting’ and investment to an external vendor’s multi-tenant virtual private Cloud environment. This not only reduces cost and overhead, but also dramatically decreases the time to market for Cloud services. An IaaS Cloud provider can rapidly set up a hosted virtual private Cloud on an enterprise’s behalf. These cloud environments are bounded by specific enterprise security parameters.
A hybrid solution also makes sense in cases where an application, or some functionality (e.g., testing environment) is hosted on a public Cloud, perhaps because it needs to be customer facing for example but it still needs to interface with some APIs that are hosted via the internal Private cloud. In such a case, the application can be spread amongst the internal Cloud infrastructure as well as on the public Cloud. This calls for Cloud interoperability, which is extremely important, given the connected environment and expanding ecosystem of innovation.
Regardless of the decision of whether to actually operate a Cloud, or source a private Cloud, there will be a host of operational implications to address – but at the end of the day, the ability to host or operate a cloud can be satisfied through capital investment and organizational change, however the decision to operate or source will be a strategic level decision. Will the business model be supported through a Cloud platform on-premise or will it be supported through a Cloud platform sourced externally – that is the question to ask.
From a technical perspective, hosting a private Cloud requires full fledged Cloud service lifecycle management. From being able to automate full-stack layered provisioning across heterogeneous platforms within the enterprise to being able to manage service retirement and resource reclamation. Investments in virtualization software, process automation, service provisioning and monitoring, infrastructure software and hardware upgrades in line with next generation data center features. Other important areas needed to support a private Cloud are the abilities to provide a self-service portal, supported by a service catalog – The enterprise will either need to develop or purchase a front end application from which users will access the cloud. In an optimal configuration, users will be able to access the cloud from inside the enterprise’s network as well as outside the network via a secure VPN session, or through a secure web browser, which is often seen in the public cloud.
A successful Cloud computing environment will require virtualized servers that interact with virtualized storage and virtualized network components in order to provide access services and data requested by users. Elastic load balancing is critical to the performance and capacity of the cloud as a pool of users will be accessing the cloud at different times for different services. In the event the pool of users changes, elastic load balancing will allow the enterprise to automatically add or remove computing resources without impacting the user’s experience.
Being able to master these aspects allow for an enterprise to potentially expand its private Cloud to the external world (i.e., offering a public Cloud service).
The question to ask is, does your enterprise have what it takes to pursue a private Cloud? This isn’t solely about the technology, but about change management, culture and organization. This calls for standardization of commonly repeated operating procedures related to the Cloud lifecycle, from provisioning to monitoring, measurement and de-commissioning. Deployment and management processes should be fully automated in order to provide agility and true elasticity. There is the need to build a services interface to the business which allows for a chargeback model that allows sophisticated measurement and chargeback capabilities to support ongoing capacity planning. From a customer service aspect, the ability to provide self service access to users entails requires sophisticated workflow automation in order to increase the speed and efficiency of Cloud deployments.
For larger enterprises, public Cloud computing is extremely viable for the commodity services with less business and data critical constraints, especially for services with ‘bursty’ characteristics – in such cases, the sheer scale of transactions, and agility allowed by procuring services from an elastic environment can justify the Public cloud approach. However, there is a point where the cost justification of hosting such services suggests hosting a private Cloud instead of procuring from a public Cloud. Even if the numbers do add up, does it make sense to invest energy into an area which the public Cloud could potentially cater to in the longer term, or where prices might decrease? Moreover, is there alignment with your organization’s corporate culture, competencies and fundamental DNA in hosting a private Cloud – the change management processes and investment needed from a cultural perspective – are they worth it?
Choosing whether to host a service on a private cloud, or procure it from a public Cloud depends on where the service lies on this spectrum. Services that are critical to differentiating the business will not be easily publically available, moreover, there are likely to be heavy compliance, security and governance restrictions which public Cloud service providers may not currently offer, though over time this may change as demand increases and public Cloud service providers evolve their service standards and agreements. In theory it could be possible to composite services and leverage the diversity of the greater ecosystem to source disparate services in a piecemeal fashion, and aggregating them through a Cloud brokering service who could potentially composite the service tailored to a particular business requirements. However this will require standardization of Cloud interoperability interfaces, metadata and other Cloud related architecture standards and protocols which all are a work in progress to say the least. Given this, such services are potential candidates for private Clouds, however there are still other factors such as economic return, technical and organizational feasibility and other risks and controls that need to be considered.
One could instinctively think that Services which are on the commodity end of the spectrum are strong candidates to be sourced from the public since they are heavily standardized and reusable. This is a fair assumption in many cases; however there will be cases where concerned data and information is business and security critical. In such cases, services would be potential private Cloud candidates.
Aside from analyzing the spectrum of services, network latency may be an issue to consider, depending on the geographic footprint of the organization concerned. Performance is paramount, especially with applications that have to send and receive a great amount of data, and in many cases the latency of the Internet is not acceptable in which case public clouds will not be a viable option.
Some argue that in reality a private Cloud won’t be able to match the elasticity of a public Cloud since it will be restricted to a smaller set of systems, especially given the risk aversion of various business unit and IT managers within the enterprise, as well as the cost benefits varying across the various business uses and IT. In theory this may be true however in practice, we don’t believe this should pose a problem if capacity planning and all other requisite process workflow automation is in place.
One drawback that should be considered is that a private Cloud infrastructure stack needs to be distributed across a heterogeneous environment of many different types of business units and systems, each having their own SLAs. So, as complexity of the environment increases, greater caution (and cost) is needed to manage the private Cloud environment and as more internal customers start to migrate towards the internal Cloud, management of the increasing demands becomes an ever more critical issue.
A key differentiator between a public Cloud and private Cloud boils down to a combination of service access, ownership and control. One can imagine a Cloud service source spectrum of Cloud service sources – from pure private Clouds, where services are internally focused on a specific enterprise all the way to pure public Clouds where services are available to all consumers. In between will be examples of shared cloud services for sprawling organizations (such as the federal government), business partners in a supply chain, enterprises in the same corporate park and noncompetitive organizations, such as universities…
There are many moving parts that go into the decision making process of which Cloud service sourcing strategy to pursue. At the most basic level, all decisions, IT and business, need to flow from the corporate strategy downwards to the business strategy level, which in turn feeds into functional and operational strategies.
This is a typical juncture where technology decisions are made in the enterprise. That is not to say that there aren’t extraneous circumstances where the corporate and business’ strategy feeds directly into technology decisions, especially in internet and hi-tech companies such as Google, VMware and Amazon.com, however on aggregate, technology in the traditional enterprise is an enabler of a functional and operational strategy and will be for the foreseeable future. The reference to ‘Technology Strategy’ in Figure 1 implies a broad spectrum of technology and IT decisions, however the focal point of this paper hinges primarily on Cloud Computing Strategy obviously.
Understand the Services is a crucial step towards formulating a Cloud Strategy –
To even being thinking about the Cloud, it is imperative to have access to a Services Portfolio, that is to be able to document and aggregate all services needed to support the functional and operational strategies of the enterprise. Depending on the complexity of your organization, it could be something as simple as an inventory list of services to the more typical and recommended approach of utilizing a portfolio management toolkit.
In many cases, Cloud services either do not exist today, are not proven, do not meet regulatory or legal requirements or simply don’t meet the specific service-level requirements that enterprises desire. Though it is likely that this will remain for some services, we feel that as the industry evolves many of these criteria will being to meet enterprise standards as Cloud computing becomes more and more mainstream.
One of the first steps in formulating a Cloud strategy is to get an idea of which services area candidates for Cloud services, and going a step further, taking an initial educated guess on whether it can be sourced from a private or public Cloud which can be done be understanding the scope and criticality of each service.
It is important to realize that not all services are suitable for Cloud computing. There is a spectrum of services, at one end of the spectrum lies the business critical services – these are measured in terms of functionality, ability to please the customer and differentiate the business. At the other end of the spectrum are the typical standard services, these can be thought of as commodity services such as basic functional offerings related to finance, accounting, IT and so on. These services aren’t highly customized or specific to a business per se.
Bottom Line – there are many moving parts and it is imperative to reconcile business and operations strategy with your technology and Cloud strategy, and to formulate your Cloud strategy be sure you understand the landscape of services your organization provides and consumes!
Cloud Computing is for real, and the burning question isn’t whether Cloud Computing should figure into IT strategic planning, but rather on the lines of how Cloud Computing fits into the IT landscape. When approaching this question, one sure thing you need to ascertain is exactly what type of Cloud model is optimal for your organization, public or private? Perhaps a blend of both?
Fundamental to being able to devise a Cloud strategy and understand which delivery model to pursue, it is important to drill down into some fundamental constraints that will power your decision. This should be approached both from a business strategy point of view, as well as from an operational efficiency perspective. Does it figure into your organization’s corporate strategy to be a consumer of cloud services, or a provider of cloud services? What are the types of services in the running to be consumed over the cloud – are they appropriate to be offered via the Cloud – are there specific regulatory and legal constraints that need to be considered? Furthermore, the design of the enterprise application architecture will need to be taken into account, as well as the maturity of your organization’s IT infrastructure. An easily overlooked aspect of ascertaining a Cloud strategy is how effectively can process be automated and governed within the organization – hosting a private Cloud is a paradigm shift, not just technologically, but from a process, organization and skills perspective. Last but not least – the bottom line – does it make financial sense to maintain Cloud infrastructure in house, or to source it externally? As you can see, deciding on what your organization’s Cloud strategy is like looking yourself in the mirror and doing some deep, realistic introspection.
In the NIST definition of Cloud Computing, virtualization and elasticity are listed as requirements. Virtualization is made possible by implementing hypervisors on servers, thereby allowing for multiple instances of operating systems to lie on a single server and sort of giving the impressions that you have many many servers available. Before this happened, you’d typically have a single operating system on a single server, and once that server is used to capacity, the only way to horizontally scale (to add capacity) would be to physically add more servers to your cluster.
Adding more servers meant you’d need to connect the other parts of the infrastructure stack, i.e. network components and storage components to the newly added servers – which can be a pain when you are scaling rapidly.
Enter, converged architectures which are integrated stacks of storage, network and servers, with virtualization hypervisors automatically built on top. Their advantage is that when you need to horizontally scale out, you just have the hypervisor automatically govern the resources of the infrastructure stack – think of the power of this for a moment…You have a single point of control which greatly simplifies how you manage your infrastructure, which by the way, results in lower operational expenditures and lower resource requirements.
Read here for a great case study.