Ask five technologists what cloud, hyperconvergence and software-defined infrastructure mean, and you will get five (or more) different answers. That’s because there are no standard definitions of these buzz words. You cannot ignore them. In fact, you need to understand these concepts so you can make the best decision for your business.
But how? I have found that the best approach may be to tell the story of the data center and how each of these trends can help balance data center risk with data center opportunities. What story you ask? Before I get started, I wanted to make one comment; there are many who have passionate, complex views on these trends but I will present my view in the simplest form. Read on.
There are three main hardware ingredients in a data center; servers, storage and networking. We all know about server virtualization. What if I told you that hyperconvergence (at its heart) is storage virtualization, and software-defined networking (at its heart) is network virtualization. So what is software-defined infrastructure? It’s the combination of virtualized servers, virtualized storage and virtualized networking. Simple? Here is the story of how all this happened.
Not too long ago, life in an x86 data center was less complex. Servers ran single workloads and storage was connected to servers and the outside world via networking. Servers were delivered by a server vendor, storage by a storage vendor and networking by a networking vendor. Very simple.
Chapter 1: Server Virtualization on Converged Systems
Why virtualize you ask? Again, it’s simple. In fact, if you are reading this on a Windows-based computer, odds are that you are utilizing less than seven percent of your system’s processing power and 50 percent of its physical memory. That feels like a lot of waste (by the way, a simple Ctrl-Alt-Delete on a Lenovo computer will quickly show you these statistics via the Task Manager). Server virtualization lets users run more workloads per server so utilization goes up and waste goes down. The majority of data centers are leveraging server virtualization as today, most workloads can be virtualized — including very advanced applications such as the SAP HANA in-memory database and analytics software.
Ok, so what are converged systems? Its ends up that delivering virtualized x86 servers with all the infrastructure they need is quite complex, and to resolve this complexity, Lenovo, IBM, Cisco, NetApp, EMC and others started putting these systems together for customers. This greatly reduced the complexity and increased adoption of virtualization. So much that in 2014, the converged systems market segment was six billion USD and growing at 33 percent compound annual growth rate (CAGR). To put that into context, traditional infrastructure growth was materially flat.
Chapter 2: Storage Virtualization on Hyperconverged Systems
While converged systems reduce the complexity and cost of virtualized servers, something else was needed. At the end of the day, sharing storage among servers is a basic virtualization requirement. This ignited the question, what if I could virtualize storage via software? And the answer, then I could save even more money while also removing complexity from my data center. A group of companies are doing just that (in different flavors). But how do I virtualize storage? And is there an ideal hardware platform for this type of storage virtualization?
That’s where hyperconverged systems come into play. Where a converged system combines virtualized servers, storage and networking in a 42U rack, hyperconverged systems can combine all that in a small 2U space (or 1/21st) of a 42U rack. Businesses can simply add as many of these hyperconverged systems together, virtualize the storage and off they go. Now that servers and storage are virtualized, data center cost and complexity is further reduced. Note that storage virtualization is sometimes also called software-defined storage. What’s next?
Chapter 3: Network Virtualization on Software-Defined Systems
The next logical choice is network virtualization. What exactly does that mean? Haven’t VMware and others already virtualized networks? The answer is yes, but only as an enabler for virtualized servers. Are there more cost savings to be had you ask? The answer is yes. Data centers acquire lots of switches — each potentially something that needs to be managed and can lead to complexity. What if you could use industry-standard switching hardware while controlling entire groups of switches from a single place (like an x86 server). This would lead to all sorts of benefits such as lower cost, better compatibility and less lock-in.
Software-defined networking technology holds that promise and is the final piece to a less complex and less costly data center. A system that has virtualized servers, virtualized storage and virtualized networking is often referred to as software-defined system. A data center-wide implementation (although none exists today) is referred to as software-defined data center.
This story is an over-simplification so as to explain IT trends and technologies in an easy way. Clearly, there are many nuances and differences on a deeper level in these technologies. That said, while the path to a software-defined data center holds real value for businesses, this is a journey. Businesses should put together a transformation plan that balances risk and opportunities as they execute these transitions. With all this transition, what remains steady is that x86 servers will continue to command greater importance and value in the data center. Your choice of the right x86 solution provider will play a key role towards a happy ending in your unique data center story. We hope you choose Lenovo systems.