IT infrastructure management has come a long way. As equipment footprints have shrunk, the capacity for information processing has grown – which means that we can do much more with less equipment. With the advent of virtualization and cloud computing, a new wave of data center innovation is upon us: superconvergence.
Toward a Unified Platform Architecture
Digital devices have traditionally been created for specific functions. These were defined during the design stage. Hardware and software specifications were crafted to enable individual equipment to perform distinctly as servers, workstations, routers, switches, firewalls, storage units, load balancers – the list goes on. The special character of each device normally meant that its capabilities were limited. It is now possible to have all these functions on a single platform.
The concept of a unified platform architecture has been studied for years. In their 2008 paper “The Architecture of Platforms: A Unified View,” Harvard Business School researchers Baldwin and Woodward wrote that “complex systems, by definition, have many parts that must work together as a whole.” They cited the Oxford definition of a platform as “a discrete structure intended for a particular activity or operation.” A unified architecture involves a collection of assets that allows for adaptability through modularization and the use of stable interfaces to connect the core components.
The Evolution of IT Infrastructure
The words of these academics may be helpful in understanding the quest of IT infrastructure designers. Professionals have been searching for years for that elusive single pane of glass, that view that allows them to see the entire IT infrastructure in one window. We are getting closer, but it has taken significant developments in levels of convergence to get us there.
In the whitepaper “The 4th Era of IT Infrastructure: Superconverged Systems,” Cloudistics Chief Scientist Dr. Jai Menon describes four generations:
- 1st Generation: Siloed
- 2nd Generation: Converged infrastructure
- 3rd Generation: Hyperconverged infrastructure
- 4th Generation: Superconverged infrastructure
Each successive generation is progressively more converged. In the 1st generation, IT infrastructure is siloed into four separate functional units: compute, storage, networking and virtualization. In the converged architecture of the 2nd generation, it became possible to manage these silos through a single software application, but physical devices remained distinct. Hyperconverged infrastructures, the 3rd generation, combined functions in a single device – usually compute and storage, but rarely including networking.
In a superconverged network, however, all of these functions are included in a single platform. The 4th generation of IT infrastructure brings together compute, storage, networking, virtualization and management, all in a tight package. One physical device does it all, providing the unified platform architecture envisioned by our Harvard researchers.
The Contours of a Superconverged Environment
So what does a superconverged environment look like? Imagine all the capabilities of the data center neatly packaged into a single physical device:
- Compute
- Storage
- Network
- Virtualization
- Management
No longer are IT functionalities divvied up among a plethora of physical devices. Everything you need is there. For example, the IT infrastructure firm Cloudistics offers a tightly integrated platform with its three-tiered Ignite device. It has been referred to as a “data center in a box.” The system includes:
- Network Block: 48x10GbE ports, 6x40GbE ports
- Storage Block: 12–112TB flash usable capacity
- Compute Block: 2–8 nodes
Single-pane-of-glass management is available from anywhere in the cloud. Each block is independently scalable, and the entire infrastructure is defined by software. To create the infrastructure, Ignite includes:
- Compute (hardware appliance)
- Hypervisor
- Storage virtualization
- Network virtualization
The days of dedicated physical boxes for each IT functionality may be nearing an end. Just as your mobile phone is quickly becoming an all-in-one device, the same is happening for the data center. Add to the mix the advanced technologies software-defined networking (SDN) and network functions virtualization (NFV), and now you have an extremely capable platform. The result is a virtual world of computing, where the physical and control dimensions of the IT infrastructure are completely decoupled.
Benefits and Challenges of Superconvergence
If convergence has its advantages, then superconvergence has more. Having everything in one box eliminates all the potential points of failure that might exist in a network of separate devices. Latency is reduced and performance is improved. The potential benefits of a cloud architecture through this “data center in a box” may seem intuitively obvious, but they deserve further elaboration. To be fair, we should also make mention of what may seem to be an almost Luddite objection to the cloud environment.
The first salient point is the virtue of simplicity of deployment. Take it out of the box, hook it up to power and the network, and you’re on your way. Try doing the same thing, for example, with four different servers, separate switches and routers, and any number of other physical devices. It will take a while to set all those up. The footprint of convergence is small, and the physical setup is easy.
Next, a superconvergent infrastructure is eminently scalable. Elastic block flash (EBF) allows use of storage capacity in configurable amounts. From a single management window, virtualized devices can be added, duplicated or removed with a few clicks of the mouse. Servers in the compute block can be allocated as necessary.
Another obvious advantage of superconvergence is its cost effectiveness. Why purchase multiple systems and licenses when you can buy just one?
One notable difference between hyperconverged infrastructure and superconvergence is the networking component. Most hyperconverged solutions leave out the network switch. Superconverged infrastructure includes network as well as virtualization and management.
Central management is a key feature of superconvergence. With a single-pane view, network managers will have a bird’s eye view of the infrastructure and will be able to drill down as needed.
With a data center in a box, it’s possible to ensure business continuity and a resilient architecture. Single points of failure are theoretically eliminated, drive protection can be implemented, and non-disruptive upgrades accomplished. Storage is elastic, and allows for easy scaling.
There are some who have voiced objections to cloud computing in a converged environment. Perceived lapses in performance have led some to return to “bare metal” because of what may be considered a virtualization tax. Reluctance to adopt new technology is nothing new, and these objections may dissolve in time.
Conclusion
Advances in technology are cumulative. Superconvergence is the culmination of decades of digital computing development. Bringing all technologies together into a neat little package has been an elusive dream for many. As more equipment providers combine functionalities into a unified platform architecture, any wrinkles in the solution will be ironed out in short order. It remains to be seen just how quickly the concept is adopted within the industry. It will likely represent a significant trend in the management of IT infrastructure for the foreseeable future.