At VMworld this year, both in San Francisco and Barcelona, VMware CEO Pat Gelsinger introduced the concept of the Software-Defined Datacenter (SDDC). This builds on the concept that as more and more of the Data Center becomes virtualized (servers, desktops), delivering greater cost-savings and agility to customers, software-defined automation and functionality (network, security, storage, backup) become the next logical steps to help IT deliver greater value to the business.
As with any new technology or vision, there are often many questions about how this will impact the market, how it will affect IT organizations. Wikibon did a nice job providing their view on "Software-led Infrastructure". It's one of many attempts that I've seen to start trying to put a scope around this concept. Some portions are agreed upon, while others are creating some headaches.
I created this short FAQ to help answer some of those questions:
1. VMware is using a new term, “Software-Defined Datacenter” (SDDC), at the center of the 2012 conference. What is Software-Defined Datacenter?
[Steve Herrod blog]. Software Defined Data Center is VMware’s vision that greater business value can be created from IT when intelligent software is abstracted from standardized hardware. In the simplest technical definition, it is the separation (or abstraction) of the “control plane” (configuration, topology awareness, management, operations) from the “data plane” (moving data, storing data).
1a. Is there a clear spelling of this term?
- Meh. Maybe, but it will have at least 3-5 variations in 2013. Just call it "SDDC" and save yourself a lot of auto-correct headaches.
- Software-Defined Datacenter is not defined by an existing standards body (eg. IETF, ITU, NIST), but rather it is vision for the evolution of how Data Center environments will become more flexible in responding to business demands. SDDC builds upon the abstraction that server virtualization has created and extends this to broader elements of the Data Center (eg. network, storage), as well as expanding the roll that automation will play in the future.
3. How is “Software-Defined Datacenter” different than “Cloud”?
- Cloud (or Cloud Computing) is fundamentally a new operational model for IT, where resources are delivered on-demand. While Cloud uses technologies such as virtualization or converged infrastructure, it’s primarily about the shift in delivery and consumption of IT services. Software Defined Data Center is the next evolution of the underlying technology, where software delivers greater levels of intelligence and value, on top of standardized hardware.
4. Does Software-Defined Datacenter eliminate the need for traditional Data Center hardware?
- No. There will still be a need for physical serves (CPU, memory), network devices to connect ports and deliver bandwidth, and devices that can store data on flash/disk/tape. But the trend in the industry is that these devices are becoming more standardized on x86 chips, mass produced memory/disks and mass produced ASICs. This trend should allow faster, more simplified “fabrics” (interconnecting servers, networks and storage) to be built, with the intelligence for policy, security, operations to continue to move into software, which is faster to develop and adapt to changing business requirements. Leading companies have been shifting their product strategies to embrace this trend for the last few years.
5. Which market segments does Software-Defined Datacenter target, or which use cases?
- Software-Defined Datacenter technology are applicable to markets of all sizes (Enterprise, Mid-Market, Service Provider), but the initial adopters have been large Service Providers that are attempting to solve challenges with large-scale Data Centers. As the competition for Public and Hybrid Cloud services increases (Amazon, Google, Rackspace, Microsoft, Cloud Service Providers), the need to drive greater operational efficiency, and associated costs and time-to-market, is pushing them to solve problems in new software-centric ways.
- As more Enterprise and Mid-Market customers adopt Private Cloud and deliver IT-as-a-Service, I also expect SDDC technologies to evolve to solve challenges at different scale, as well as user-centric challenges such as BYOD.
- Even more than ever, the current era of IT is ultimately defined by rapid change, in terms of new devices (smartphones, tablets), new application consumption models (PaaS, SaaS), or converging technology silos (virtualization, converged infrastructure). Software-Defined Datacenter is the next step in converging functional areas, while attempting to give IT the ability to respond to business challenges faster.
- As mentioned above, Software-Defined Datacenter does not eliminate the need for physical hardware within the Data Center. Rather it is a vision to enable customers to better take advantage of the trend towards delivering software intelligence on standardized hardware. As with many technology transitions, there are opportunities to evolve technology portfolios, evolve business models and unlock new partnership opportunities.
8. Is Software-Defined Datacenter explicitly linked with open-source technologies such as OpenStack, OpenFlow or Open vSwitch?
- While there are open-source projects today that will have an influence on Software-Defined Datacenters, by no means does this mean that this is the only delivery mechanism for customers to obtain the technology needed for this IT technology evolution. A few examples of this:
- VMware’s acquisition of Nicira – while Nicira was a major contributor to the OpenStack Quantum project (network virtualization) and the Open vSwitch project, which are both open-source, their core NVP product was a commercial offering.
- OpenFlow is a standards-based protocol for network virtualization that can be implemented by any vendor, for either open-source or commercial products.
- “Project Razor”is an open-source project that was jointly created by EMC and Puppet Labs to deliver advanced server and application automation for Data Center and Cloud environments. The software can be used with either commercial products (eg. VMware vSphere, Cisco UCS, etc.) or open-source projects (OpenStack, KVM, CloudFoundry)