I was having a discussion with a skeptical CIO the other day. His issue was that a private cloud isn’t real. Why? In contrast to the public cloud, which has unlimited capability on demand, a private cloud is limited by the size and capacity of the internal data center. While I understand this point I disagree and here is why. I don’t know of any data center that doesn’t have enough servers or capacity. In fact, if you talk to most IT managers they will quickly admit that they don’t lack physical resources. This is why there has been so much focus on server virtualization. With server virtualization, these organizations actually get rid of servers and make their IT organization more efficient.
Even when data centers are able to improve their efficiency, they still do not lack resources. What data centers lack is the organizational structure to enable provisioning of those resources in a proactive and efficient way. The converse is also true: data centers lack the ability to reclaim resources once they have been provisioned.
So, I maintain that the problem with the data center is not a lack of resources but rather the management and the automation of those resources. Imagine an organization leverages the existing physical resources in a data center by adding self-service provisioning and business process rules for allocating resources based on business need. This would mean that when developers start working on a project they are allocated the amount of resources they need – not what they want. More importantly, when the project is over, those resources are returned to the pool.
This, of course, does not work for every application and every workload in the data center. There are applications that are highly specialized and are not going to benefit from automation. However, there indeed can increasingly large aspects of computing that can be transformed in the private cloud environment based on truly tuning workloads and resources to make the private cloud as elastic as what we think of as a ever expanding public cloud.
To comprehend HP’s cloud computing strategy you have to first understand HP’s Matrix Blade System. HP announced the Matrix system in April of 2009 as a prepackaged fabric-based system. Because Matrix was designed as a packaged environment, it has become the lynch pin of HP’s cloud strategy.
So, what is Matrix? Within this environment, HP has pre-integrated servers, networking, storage, and software (primarily orchestration to customize workflow). In essence, Matrix is a Unified Computing System so that it supports both physical blades as well as virtual configurations. It includes a graphical command center console to manage resource pools, physical and virtual servers and network connectivity. On the software side, Matrix provides an abstraction layer that supports workload provisioning and workflow based policy management that can determine where workloads will run. The environment supports the VMware hypervisor, open source KVM, and Microsoft’s Hyper-V.
HP’s strategy is to combine this Matrix system, which it has positioned as its private cloud, with a public compute cloud. In addition, HP is incorporating its lifecycle management software and its security acquisitions as part of its overall cloud strategy. It is leveraging the HP services (formerly EDS) to offer a hosted private cloud and traditional outsourcing as part of an overall plan. HP is hoping to leveraging its services expertise in running large enterprise packaged software
There are three components to the HP cloud strategy:
- Cloud Services Automation
- Cloud Consulting Services
CloudSystem. What HP calls CloudSystem is, in fact, based on the Matrix blade system. The Matrix Blade System uses a common rack enclosure to support all the blades produced by HP. The Matrix is a packaging of is what HP calls an operating environment that includes provisioning software, virtualization, a self-service portal and management tools to manage resources pools. HP considers its public cloud services to be part of the CloudSystem. To provide a hybrid cloud computing environment, HP will offer compute public cloud services similar to what is available from Amazon EC2. When combined with the outsourcing services from HP Services, HP contends that it provides a common architectural framework across public, private, virtualized servers, and outsourcing. It includes what HP is calling cloud maps. Cloud maps are configuration templates based on HP’s acquisition of Stratavia, a database and application automation software company.
Cloud Service Automation. The CloudSystem is intended to make use of Services Automation software called Cloud Service Automation (CSA). The components of CSA include a self-service portal that manages a service catalog. The service catalog describes each service that is intended to be used as part of the cloud environment. Within the catalog, the required service level is defined. In addition, the CSA can meter the use of services and can provide visibility to the performance of each service. A second capability is a cloud controller, based on the orchestration technology from HP’s Opsware acquisition. A third component, the resource manager provide provisioning and monitoring services. The objective of CSA is to provide end-to-end lifecycle management of the CloudSystem.
Cloud Consulting Services. HP is taking advantage of EDS’s experience in managing computing infrastructure as the foundation for its cloud consulting services offerings. HP also leverages its consulting services that were traditionally part of HP as well as services from EDS. Therefore, HP has deep experience in designing and running Cloud seminars and strategy engagements for customers.
From HP’s perspective, it is taking a hybrid approach to cloud computing. What does HP mean by Hybrid? Basically, HP’s hybrid strategy includes the combination of the CloudSystem – a hardware-based private cloud, its own public compute services, and traditional outsourcing.
The Bottom Line. Making the transition to becoming a major cloud computing vendor is complicated. The market is young and still in transition. HP has many interesting building blocks that have the potential to make it an important player. Leveraging the Matrix Blade System is a pragmatic move since it is already an integrated and highly abstracted platform. However, it will have to provide more services that increase the ability of its customers to use the CloudSystem to create an elastic and flexible computing platform. The Cloud Automation Services is a good start but still requires more evolution. For example, it needs to add more capabilities into its service catalog. Leveraging its Systinet registry/repository as part of its service catalog would be advisable. I also think that HP needs to package its security offerings to be cloud specific. This includes both in the governance and compliance area as well as Identity Management.
Just how much will HP plan to compete in the public cloud space is uncertain. Can HP be effective in both markets? Does it need to combine its offerings or create two different business models?
It is clear that HP wants to make cloud computing the cornerstone of its “Instant-On Enterprise” strategy announced last year. In essence, Instant-on Enterprise is intended to make it easier for customers to consume data center capabilities including infrastructure, applications, and services. This is a good vision in keeping with what customers need. And plainly cloud computing is an essential ingredient in achieving this ambitious strategy.