How do you treat hardware like software? That question sounds like a contradiction, but we've been helping customers answer this question for the past six years with Cisco UCS. When you abstract all configuration and identity of hardware and transform it into software defined infrastructure (SDI), or better yet,policy driveninfrastructure, you're moving down the path of managing the "infrastructure as code."

An essential aspect of this automated management is encapsulating the best practices of your server, storage and network experts as policies and templates. Cisco describes these as Service Profiles. The Service Profiles combined with the open Application Program Interface (API) in UCS provide a common "language' for provisioning and configuring the infrastructure across the different types of devices. As we examined in a previous blog in this series, the combination of true SDI plus best practices defined in Service Profiles makes sure routine tasks are implemented consistently and correctly to reduce risk. Our customers are receiving tremendous benefits using Service-Profiles today with their existing UCS blade and rack systems, and we have extended this same management framework to our composable infrastructure.

Here's where it gets fun: DevOps and Infrastructure as Code

DevOps is a growing movement toward processes and methods for communication and collaboration between development, QA, and IT operations. It requires establishment of standard ways of provisioning and configuring infrastructure in a programmatic manner.  Infrastructure as code implies that the infrastructure resources can be composed in a consistent fashion based on the demands of the application. Cisco UCS offers Service Profiles and a fully open and documented API to accomplish this.

In our recent discussion with IDC analyst, Jed Scaramella, we asked Jed to describe the role of policies and templates and the benefits for customers. Jed explained in this video that the disaggregation of servers with composable infrastructure adds complexity by creating more elements that must be managed. Achieving greater efficiency and speed requires more simplicity. Service Profiles provide this simplicity by defining a consistent way of managing infrastructure resources through software. They optimize the resources by allowing  applications and infrastructure to be deployed consistently at a faster rate while reducing risks. That's why Service Profiles are essential to effectively implementing composable infrastructure.

Precursors to Composable Infrastructure

Composable infrastructure is predicated on two key capabilities:  (1) disaggregation to decompose systems (primarily servers) into subsystem elements and (2) a control plane complete with an open API to reassemble the pieces.   This allows the physical and logical infrastructure to be made available to orchestration tools as granular pools of resources.   An important aside:  traditional blade and rack servers are out of the equation...they're still monolithic, fixed servers that you can only right-size at the factory or carve up with a hypervisor or containers in order to optimize.  Or said another way, you can't compose a system out of HW that is already fixed as a system. If someone tries to sell you yet another blade server as composable infrastructure they're "composable washing." Brace yourself, because it's going to happen a lot now that everyone is casting around for the next big buzzword to follow "hyperconverged."   It's an ugly business, marketing.

UCS: Composable Infrastructure, AKA Infrastructure as Code, AKA Bad to the Bone if you like to automate

Cisco System Link Technology achieves server disaggregation and provides the control plane functionality to combine server subsystems together (processor/memory, local or network storage and I/O.)    UCS Service Profiles provide the templates that define the servers you want to create out of these resources.    This is a pretty stunning advancement:  physical servers that can be dynamically composed and optimized for the unique requirements of workload.

By managing the infrastructure as code, you can deploy applications faster by rapidly composing the infrastructure you need in a consistent and predictable way. Figure 2 below illustrates how templates allocate the resources from the pool, then they are provisioned and configured (composed) to uniquely support each application and workload. This is composable infrastructure in action and gives your data center infrastructure powerful public cloud-like capabilities.

Fig. 2: Templates in UCS Management Allocate and Compose Infrastructure Resources on Demand

As Jim described in a blog when we announced our first composable product, the UCS M-Series,  "With the policy-based provisioning, deployment and management of UCS, the application can be presented the exact amount of resources from each subsystem for truly a software-defined server."

Step Away From the Forklift

Most organizations currently use VMware vCenter, Microsoft System Center and other tools in their daily operation. They don't plan to implement new tools just to support composable infrastructure or adopt DevOps. They shouldn't have to. UCS management includes a broad, mature partner ecosystem that includes deep integrations with these tools and many others. This reduces risk by allowing IT teams to use existing management tools and processes as they incorporate new technologies, like composable infrastructure, or implement new methodologies like DevOps.

Figure 2, below, shows one example of how Cisco enables the implementation of DevOps and composable infrastructure in a way that works best for your organization. UCS management supports the entire UCS product portfolio, including Cisco Composable Infrastructure such as M-Series and the UCS C3260. Cisco offers a vSphere plug-in that we recently updated; it enables users to work from their familiar VMware tools. Just as VMware has worked with Cisco and other vendors to develop plug-ins, they also have also worked with Chef to develop a Chef knife vSphere integration. The Chef integration enables users to provision, list, clone, and delete VMs managed with vCenter along with other common tasks from their familiar vSphere client. These are capabilities that support advanced automation and DevOps workflows in a simple and flexible manner.

Figure 2: Example of Chef knife and UCS Management integrations with vSphere

You don't have to wait for the future. You can start on the path to DevOps and composable infrastructure today using Cisco UCS and industry standard tools. Cisco has nearly 50,000 customers that use the advanced automation of UCS to achieve greater efficiencies in their organizations every day. Our management software is mature and proven, and it is supported by a robust partner ecosystem. Cisco Composable Infrastructure is simply another advancement that allows for automation at a much more granular level, all the way to server subsystems.

Cisco is offering more than a vision. We're offering products and real business benefits today and a path to the future built on a solid foundation.

More to come,

Todd

For additional information:

• Download the IDC paper on composable infrastructure
• Go to www.cisco.com/go/composable