Logical Abstraction of the Physical Data Center
Logical abstraction can now be applied to the entire data center, and the virtual layer can now support enterprise functionality
February 24, 2015
![Logical Abstraction of the Physical Data Center Logical Abstraction of the Physical Data Center](https://eu-images.contentstack.com/v3/assets/blt8eb3cdfc1fce5194/bltf94e95b5fe89a16d/66211e0f7144fdb5901726ef/datacenters-inside-e1421770068531.jpg?width=1280&auto=webp&quality=95&format=jpg&disable=upscale)
We’ve got virtualization, cloud computing, and now quite a bit of new buzz around commodity platforms. The reality is that this computing model has taken off faster than really expected. With the help of the modern hypervisor, software-defined technologies (SDx), the powerful systems designed to optimize your data center and allow your cloud to be a lot more agile can potentially run on commodity gear.
Here’s the reality: it’s making some traditional technologies a bit nervous.
I recently wrote that the conversation around custom-build servers, networking components and now storage has been heating up. The concept of a commodity data center is no longer locked away for mega-data centers or large organizations. Looking at Google as an example, you’ve got an organization which builds its own server platform by the thousands. In fact, Google has developed a motherboard using POWER8 server technology from IBM, and just recently showed it off at the IBM Impact 2014 conference in Las Vegas. DCK’s Rich Miller recently outlined how “POWER could represent an alternative to chips from Intel, which is believed to provide the motherboards for Google’s servers.”
Let’s start here: The logical abstraction of hardware and their associated services is the natural progression of the technological landscape.
That said, how will this impact existing physical systems? Let’s look at the components behind a commodity platform and what can be logically abstracted from the physical environment.
Data center. Say hello to the logical data center. The abstraction of data center services revolves around visibility and control ranging from the chip to the cooler… and everything in between. This “data center operating system” will be able to intelligently control a truly distributed plane. Some of the leading data center providers are already doing this. IO and their IO.OS platforms gives administrators next-gen DCIM capabilities within their data centers. Furthermore, we are seeing developments from hypervisor makers, like VMware, who aim to take control of all resources within their cluster. From there, VMware can control all underlying policies and even push workloads into the cloud.
Storage.There had to come a point where we needed to stop simply adding disks. We needed to make storage smarter. Physical storage vendors need to see the writing in the wall and adapt quickly. Already some are introducing powerful, agnostic, logical storage layers capable of direct cloud and data center interconnectivity. Let me put it this way: what’s stopping a shop from buying their own chassis and populating it with the disk of their choice? From there, point the resources to a virtual appliance like Atlantis USX or VMware vSAN and apply all of the enterprise features from there. This includes HA, encryption, deduplication, replication, cloud API extensions and more. This kind of story is becoming a lot more compelling.
Security. Virtual appliances, services, and other abstracted features are making their way into the data center. Pretty much all major security technologies now offer a physical and virtual appliance for you to work with. For example, security products from Checkpoint allow you to deploy virtual appliances and services throughout your cloud and localized network infrastructure. These software blades can be deployed on any virtualized system and include Firewall, VPN, IPS, Application Control, URL Filtering, Antivirus, Anti-Bot, Identity Awareness and Mobile Access capabilities.
Networking. Not much explanation needed here; SDN is a huge component of cloud computing and the modern data center. But that’s not what we’re talking about today. This little section is all about open source networking. Let me give you an example: Cumulus Networks has its own Linux distribution, Cumulus Linux, which was is designed to run on top of industry-standard networking hardware. Basically, it’s a software-only solution that provides the ultimate flexibility for modern data center networking designs and operations with a standard operating system, Linux. Now, go out there and build your own enterprise network infrastructure: all on open source networking.
Compute. Hardware and service profiles allow large organizations to create “follow-the-sun” data center models. These logical tools are becoming more powerful and much more automated. Hypervisor and virtualization technologies are becoming so thin, that differences in performance are almost negligible. The VMware, Hyper-V, and the latest XenServer virtualization platform continues to be an example of this. We can now point every compute resource to a hypervisor and let it take over. We can create powerful HA policies that will mirror VMs across a variety of platforms (which can be commodity). Yes, you can buy a SuperMicro chassis and create your own data center around it. However, one of the biggest concerns was always around parts, warranties, and maintenance. Now, big compute vendors are creating a “commodity-like” server line without the bells and whistles. The HP Cloudline server models, for example, will be cloud-ready without added software features while still carrying a warranty.
Taking the future into consideration. Our homes are becoming smarter, we have so much more connectivity into the cloud, and the number of users/devices coming online is growing very rapidly. Future technologies will need to take the user, the cloud, and a variety of delivery methods into consideration. In many cases, new types of logical technologies will be used. Consider these stats from the latest Cisco Visual Networking Index:
In 2013, the number of mobile-connected tablets increased 2.2-fold to 92 million, and each tablet generated 2.6 times more traffic than the average smartphone.
By the end of 2014, the number of mobile-connected devices will exceed the number of people on earth, and by 2018 there will be nearly 1.4 mobile devices per capita.
By 2018, more than half of all traffic from mobile-connected devices (almost 17 exabytes) will be offloaded to the fixed network by means of Wi-Fi devices and femtocells each month.
By 2018, over half of all devices connected to the mobile network will be “smart” devices.
We are in the midst of the Internet era as more devices, people, and data points come online. There will be new services, new kinds of compute models and new ways to deliver rich content and data. The drive of the user and the IoT will create new complex challenges around resource control. For now, many organizations are beginning to explore commodity systems paired with open-source computing, virtualization, and software-defined technologies.
About the Author
You May Also Like