Enterprise data center architecture is in the midst of the most dramatic change in decades. Hyperscale data centers have pioneered the software-defined data center (SDDC) on industry-standard hardware. Enterprise data center operators recognize the value of this approach, but few have implemented it at the network and storage layers. A software-defined data center extends the value provided by server virtualization by eliminating the remaining resource silos, increasing utilization and agility, and giving the flexibility to dynamically choose the best location and infrastructure for each workload.
Early adopters are now deploying key elements of the software-defined data center in large production environments. These products are expected to move into mainstream use by 2017. In this slideshow, Bill Stevenson, executive chairman for Sanbolic, discusses the move toward hyperconverged infrastructure and the core components of the 2017 data center.
The Enterprise Data Center of 2017
Click through for more on the move toward hyperconverged infrastructure and the core components of the 2017 data center, as identified by Bill Stevenson, executive chairman for Sanbolic.
Moving Network and Storage Services into Software
Most new storage capacity today is still purchased as expensive, inflexible storage appliances based on outdated designs. Prospectively, data centers will buy standard servers to run applications, as well as network and storage services. Storage drives and network interface components will be added based on workload requirements. A software layer running across multiple server instances will deliver the services currently provided by dedicated storage and network appliances.
The design point for this new software is to enable application elasticity, availability and user proximity, and to consistently meet service-level targets, rather than just manage the disks or network interfaces within the appliance.
Industry-Standard Hardware
Thanks to SDDC technology, hardware is becoming entirely abstracted – moving existing infrastructure into centrally managed resource pools, improving utilization and performance management, and reducing operating complexity. It also provides the capability to weave industry-standard components into the infrastructure.
Data center operators can now quickly incorporate new components into their infrastructure to benefit from designs with improved performance and lower cost, while matching workload requirements with hardware performance and cost. For example, highly transactional workloads requiring low latency can utilize flash storage and 40GbE networking, while archiving systems can utilize large-capacity disk drives and cheaper networking components.
Unified Management and Orchestration
Server virtualization improves data center flexibility – applications are no longer tied to a single physical server and can be migrated across servers within a resource pool to enable availability and load balancing. However, data is typically still tied to a single storage appliance, and each copy of the data is accessible to only one server at a time. Eliminating these constraints using software-defined storage and networking will enable truly fluid IT infrastructure.
SDDC technology enables data to be moved to where it is being used and to be shared by multiple application instances, while performance is dynamically tuned to the requirements of each application. The focus moves from managing hardware to managing the end-user experience with the application – unifying management and automating data orchestration.
Distributed Workloads
Organizations are often spread across many geographic locations, and customers are spread around the globe. While hardware-defined infrastructure is typically tied to one location, the software-defined data center of 2017 will abstract heterogeneous legacy proprietary hardware and newer industry-standard hardware into a single resource pool. Better yet, this resource pool will not be limited to one location — it can span multiple locations and/or public cloud resources as an active-active cluster managed as a single system.
By 2017, workloads will be dynamically managed across distributed resources based on factors including end-user proximity, capacity utilization, and energy cost or availability.
Infrastructure as a Service (IaaS)
Infrastructure will become an efficient, agile, elastic service for delivering the application workloads that a business depends on. Delivering the availability, performance and features required to consistently meet end-user, service-level expectations, the infrastructure of 2017 will significantly reduce the cost premium that is still common for highly available, high-performance, scalable systems.
While this has been the holy grail of IT for decades, by 2017 it will be mainstream. The economic and business agility benefits it provides are as compelling as server virtualization, which saw very rapid adoption over the past decade.
Application-Centric Infrastructure
The SDDC is much more application centric than the legacy data centers shackled by hardware-defined infrastructure and proprietary appliances. Say goodbye to the days of mapping storage capacity and network bandwidth to an application as a resource silo. Resource pools will be tuned for each application or user and be dynamically provisioned to meet performance, availability, elasticity or user proximity requirements.
There is huge value for modern, dynamic business environments in enabling new applications to be provisioned quickly and then given more or less resources in response to fluctuations in demand for the services provided by the application.
Energy Management
Energy is one of the top costs of data center operation. The 2017 data center will see accelerated adoption of energy-efficient flash storage. By 2017, more data centers will be co-locating with power plants to utilize their inherent redundancy, eliminate transmission cost and avoid the need for backup generators.
In addition, data centers will be more closely integrated with the electricity grid as part of a “smart grid.”
Power cost will become one of the determinants of where to place a workload across geo-distributed infrastructure, arbitraging the fluctuations in local power cost driven by electricity demand patterns and the variations in supply introduced by greater dependence on solar and wind power.
Don’t Rip and Replace
Migration to the data center architecture of 2017 does not require an expensive and risky strategy of ripping out racks of existing infrastructure and replacing it with new, hyperconverged appliances. Some SDDC platforms can abstract legacy hardware from multiple vendors into a resource pool with common software-defined, tier-one storage services to improve utility. It can be done swiftly for immediate benefits or incrementally, gradually moving storage services off of legacy platforms into the software-defined management layer.
Evolve Organizational Design
Significant organizational adjustments will be made to reflect the new data center. It will require new skill sets and different vendor relationships, and this can seem threatening to the existing team. Decisions are becoming more solution-focused and by 2017 the traditional model of the storage team talking to storage vendors and the networking team talking to network vendors will be long forgotten. Internal IT organizations must realign to reflect the delivery of infrastructure as a service.
The old support model provided by high-margin, proprietary IT appliance vendors will be severed, creating an opportunity for system integrators to help accelerate the introduction of SDDC architecture.
Preparing for Tomorrow … Today
The enterprise data center of 2017 will grow up to reflect many of the benefits of today’s hyperscale data centers. The dramatic cost reductions expected will be a huge win for business, but the agility of the infrastructure to efficiently deliver IT services to end users will be paramount.
The products to enable this transition are already available and can be deployed on existing infrastructure to provide a gradual migration path onto industry-standard hardware. If you start today and progress at an appropriate pace, reflecting both the organization’s capabilities and its confidence in the modern platforms, you will be better prepared for what’s ahead.