Defining the New Data Center Operating System
August 16th, 2013 By: Bill Kleyman
The data center operating system – many times referred to as DCOS. Kind of sounds like a bit of a buzz term doesn’t it? Well, the reality is that there is already a lot of conversation around the idea of an intelligent data center management platform. This isn’t a conversation about data center virtualization, but rather a look at how the future data center environment will be managed.
The modern data center is at the center of any major organization. We have a lot more systems that are being placed within the data center environment and new technologies are being born directly within a cloud infrastructure. With this focus on the data center – there needed to be a way to better manage all of the services we are relying upon. Remember, at the data center level we are still using physical components – servers, cooling equipment and advanced types of power distribution technologies. The big difference is that the physical layer is now designed around high levels of efficiency, multi-tenancy and resiliency. This is where the DCOS platform is able to take the entire data center layer and begin to abstract it.
So, how does the data center operating system really different from other DC management platforms? DCOS really means connecting the logical with the physical. As the modern data center continues to become the “data center of everything” we need to evolve the way we control these vital platforms. New types of threats against the data center environment are continuing to grow. So, to combat new security needs, an ever-scaling environment, and a lot more cloud computing; we must take the next leap in data center control and management.
- The data center operating system. Big shops like IO are getting the idea quickly. They released their IO.OS where administrators have granular visibility into components that they wouldn’t have been looking at otherwise. Furthermore, they are able to connect multiple data center nodes to create one logical cluster. This means one pane of glass – and a lot more control over the entire data center management process. They’re not the only ones getting this idea. More data center management solutions now look at the actual workloads running within the data center and how it can all be optimized and controlled.
- Connecting the physical and the logical. In the past, physical components would be separate from the virtual or logical workloads running on top. Furthermore, the management platforms would be segmented as well. The modern data center has come so far that this approach no longer applies. Both components are vital to the entire delivery process and there needs to be visibility into everything. This means seeing full metrics around both physical and non-physical systems running within the data center.
- New security trends. Advanced persistent threats (APTs) are wreaking havoc on data center workloads. Remember, these aren’t just your typical DDoS attacks. These are new, constantly live, and always present threats against physical and virtual data center components. Remember, APTs don’t just attack the workload layer of the data center. These new threats can actually take aim at physical data center components as well. Intelligent data center operating systems can now look for anomalies in normal operations and actually re-route network traffic if there is a situation. Furthermore, they are able to route this traffic at a physical and virtual layer. These DCOS platforms logically connect the data center to the applications and data that it houses. That way, it can better protect the entire logical and physical stack.
- Creating the mobile data center. No, your data center isn’t going to get up and move around (unless you have a mobile modular data center, of course.) Rather, the data center operating system presents new ways to control your infrastructure. With so much DC distribution, there needs to be a way to see multiple nodes and control them from a mobile perspective. So, in the sense of data center operating systems and management, administrators are able to control data centers from remote locations or intelligently bring multiple nodes together under one management UI. And yes, this does mean having some controls from mobile and web-based devices as well.
The abstraction of the hardware layer within the data center means we are able to utilize new types of controls around our data center. New technologies are being deployed within the data center at a very fast pace. Segmenting physical components from the workloads that are running on top is no longer the right way to approach data center management. This is why DCOS has become a powerful way of seeing all of the core and vital components running within a data center. Already, large data center providers are approaching this management methodology. What are the results? Greater visibility, more control and the ability to be proactive with very important workloads.
Couldn’t agree more Bill. While the concept of DCOS I think is foreign to most in the enterprise, those in fast moving tech companies have either already adopted solutions like IO.OS or built there own (Google/Amazon/etc.). And while all this virtualization and cloud technology we’re being enamored with every day is great and helps companies save significant cost in the datacenter,, it is also constantly adding new layers into the stack that require their own points of management and introduce their own elements of risk relating to reliability, management, and security. Having a more intelligent “DCOS” that can be aware of all those layers and intelligently change the once-static rules that defined the datacenter really makes a lot of sense. While big virtualization players like VMWare, Microsoft, and Citrix continue to add more functionality in their core hyper visor and systems management products, they will never be able to keep up with the rate of innovation that exists now in the era of the Software Defined Datacenter.
Great read, excellent points, and looking forward to seeing more development in this space.
Connecting the physical to the logical/virtual makes a ton of sense. The various touch points in the physical world for sense and control is the first half of the adventure, the ORCHESTRATION and Policy engine that drive what to DO with all that capability will evolve into a cottage business over time. The Best Practices of running a data center will be translated into RULES, and these rules will populate the Orchestration and policy engines.
Interesting side note here: The definition of the data center will expand to include all FOUR styles of computing; In-House, Co-Lo, Modular and Cloud. As thios hybrid model matures, the importance of the sense and control layer discussed above will increase.
I can hardly wait!