The Road to Software-Defined Data Center

New logical systems are capable of introducing amazing optimizations for the data center platform as well as your business model.

Bill Kleyman

August 28, 2014

5 Min Read
The Road to Software-Defined Data Center

New virtual services are sweeping the modern data center and are exciting data center administrators. Just imagine having the ability to abstract vast amounts of resources and manage heterogeneous environments all from one logical controller. New infrastructure components no longer care what type of hardware you’re using. It cares about how you are presenting resources. Software-defined technologies have come quite a long way. These logical systems are capable of introducing amazing optimizations for the data center platform as well as your business model.

But here’s the big question.How do you get there? What do you need to deploy to get to the SDDC state? The good news is some of the latest technologies out there surrounding network, storage and compute and now making it easier to become a software-defined data center.

  • Storage. Software-defined storage is a very real concept and technology. There are a number of solutions now allowing you to completely abstract storage resource and point them to a virtual layer. Let’s pretend like you’re running on a VMware vSphere hypervisor. Now, you can deploy VMware’s Virtual SAN which pools resources and allows you to create a persistent storage tier at the hypervisor layer. Or, you can even utilize a 3rd party technology like Atlantis USX and integrate it with your hypervisor to pool SAN, NAS, RAM and any type of DAS (SSD, Flash, SAS).

  • Network. The way we process and control network traffic has really come a long way. Now, we’re able to scale massive data center points with complex network routes and policies. The incorporation of software-defined networking (SDN) brings the cloud and data center conversation to a new level. VMware’s NSX solution, for example, allows you to provision, snapshot, delete and restore complex networks. The cool factor is that it can integrate with your existing network architecture. Ultimately, this gives you complete network control from a VMware management platform. Cisco’s Application Centric Infrastructure (ACI) takes the concept of SDN to a whole new level. If you’re running a Cisco architecture, you suddenly have the capability to create application-level policies for both virtual and physical workloads. Furthermore, you create automated security policies which can span logical, physical and cloud networks.

  • Compute. This is an interesting stopping point. How can you software-define the compute layer? Well, you’ll always have the hardware aspect. The big difference revolves around how the resources living on the hardware are delivered. Cisco’s UCS platform comes very close to creating a more commodity-like platform. The idea with UCS is to create granular services and hardware profiles. Essentially, you’re making blades and rack-mount servers interchangeable pieces within the architecture. You can have a chassis working until 8 p.m., at which point a policy sets in to provision a new hardware profile to support and incoming set of new users from a different time zone. From an administrative perspective, nothing really needed to happen. Hardware was dynamically reassigned and the compute layer was powered the software-defined policies.

  • Data center. VMware has taken an interesting approach to creating a true software-defined data center. Their logic is to aggregate all data center processes and allow the hypervisor and appropriate management consoles to control resources. Through network, storage, compute and management layers, the concept of the SDDC model allows administrators to completely control all aspects of a next-generation cloud data center. From within the management console, SDDC operations can include automated management, policy-driven services, better business-aware technology controls. The interesting piece here is that through VMware’s platform you’re capable of seamlessly expanding this from a private cloud into a hybrid cloud environment.

  • Cloud automation and management. There are several technologies which allow you to take your data center platform and expand it easily into the cloud. Oftentimes, data center platforms have very heterogeneous systems supporting complex application and business processes. Automation and management systems like OpenStack, CloudStack and Eucalyptus each have a way of seamlessly allowing you to expand data center resources into the cloud. These control layers allow the management of virtual machines, data points, and entire IaaS models for your organization. This is the software layer which truly allows you to manage your cloud architecture.

Putting it all together

Here’s the reality – there’s no one recipe to become a software-defined data center. Bits and pieces of virtual services and code can be deployed in unison or separately to achieve optimal data center performance. Many organizations are taking the leisurely stroll during their path to a more logically controlled data center. The beauty here is that these technologies allow traditional data center technologies to live in parallel with next-generation SDDC platforms. In some cases, it’s smart to start with storage or just networking. Identify specific points of need within your organization and begin to apply SDDC technologies. You’ll quickly notice that management becomes simplified and you suddenly regain control over quite a few resources. Furthermore, with software-defined technologies – you’re able to interconnect with various cloud models a lot easier.

Data center technologies will continue to evolve as hardware and software platforms become more interconnected. The logical aspect of the data center allows modern organizations to truly span their environments and connect with a variety of cloud resources. It’s no wonder that they hybrid cloud model is becoming so prevalent. New services allow for the powerful expansion of private data centers directly into the cloud model. All of this is done at the virtual control layer. When you begin to put network, storage and compute all together on a virtual plane – you begin to truly see just how far you’ll be able to take your own data center platform.

About the Author(s)

Bill Kleyman

Bill Kleyman has more than 15 years of experience in enterprise technology. He also enjoys writing, blogging, and educating colleagues about tech. His published and referenced work can be found on Data Center Knowledge, AFCOM, ITPro Today, InformationWeek, NetworkComputing, TechTarget, DarkReading, Forbes, CBS Interactive, Slashdot, and more.

Subscribe to the Data Center Knowledge Newsletter
Get analysis and expert insight on the latest in data center business and technology delivered to your inbox daily.

You May Also Like