Is the Concept of the 'Server' Obsolete, or in Need of Redefining?

Broadening cloud and fabric requirements are forcing redefinition of the server, and with it, the necessity for a new procurement plan and process that recognizes servers not as fixed and finite objects, but as virtualized resources and mobile instances.

• The traditional concept of the role-specific server will become increasingly irrelevant in fabric infrastructures, cloud computing and network-based virtual resources, forcing IT planners to change and adapt to the resource life cycles of integrated systems.
• By 2015, one-third of all servers will ship as managed resources integrated in a converged infrastructure, enabling providers to extract integration premiums from traditional component line-item pricing, thus creating a more complex procurement process.
• The increasing diversity of functional resources and the continual advance of mobile virtualization will force enterprises to implement more advanced levels of resource pool management and automation, which will create new competitive comparisons, alternative metrics and pricing changes that differ from traditional server price/performance benchmarks.

• Begin revising internal server road map plans from single function, separately configured systems to hybrids of storage, networks and servers.
• Reconfigure cost models and bill of materials (BOMs) of server acquisitions as part of resource pools composed of single and multinode network instances.
• Make new calculations on life cycles, resource scalability, elasticity, virtual machine (VM) mobility and placement optimization that emphasize the dynamic nature of the network versus the static generational improvements of the server.
• When purchasing integrated systems, request at least two vendor bids and request vendors to quantify promised operational productivity and management improvements.
• Request vendors of integrated systems to break out component costs in line items of server, storage, management and application software to determine logical integration premiums.
• Weight intelligent automation across multinode networks as the new competitive differentiator versus static hardware configurations.
• Procurement specialists should expand evaluations to cover integrated and managed resources (cores, memory, input/output [I/O]).

Server terminology is in need of revisions as data centers evolve from silos and isolated hosts to fabric infrastructures, integrated and converged infrastructures, cloud and virtualization. The network node represents a more modern concept of server. The node refers to a "resource" situated on a network at some location and in some form factor available to applications or workloads for execution. The resource can comprise a physical array of processor cores, memory, I/O interfaces, graphic and numeric processing elements, encryption processing and storage. Its form factor can be a blade, racks, skinless, or other configurations comprising motherboards and associated processing, storage and I/O elements. Or the resource could be a logical aggregation of these components and be represented by the terminology: multinode resources, distributed locally or geographically over networks in enterprise data centers or cloud hosting sites. The multinode resource can be a shared resource of multiple nodes (of cores and memory, etc.) defined by workload domains, which in turn reside in racks on a network. The logical aggregation of the resources is called a pool and the manager of the resources across a fabric is called the fabric resource pool manager (FRPM) by Gartner.

The instantiation of servers as nodes, and nodes as resources attached to resource pools, creates the modernized fabric infrastructure. This perspective emphasizes resources as managed entities, or instances, that are logically integrated and available to applications through aggregation and sharing, most often abstracted from the hardware as VMs. Thus, the concept of the server is becoming outdated – though server as an object will remain for years a definitional convenience in the procurement process. Users should understand that the server concept in a modernized data center denotes oneness of identity and role specificity. Contrast this to the broader requirements of cloud and fabric which imply that applications should not care about size or scale of computing, place of execution, bounded configuration, or physical housing.

The reason for the evolution in how we think about servers is due to the increase in server processing. Servers now include many other functions and features traditionally performed by dedicated and specialist devices. This relates to the rise of intelligent management, virtualization, improved storage features and network connectivity. These enhancements will drive real-time business demands of workloads, near-instant resource elasticity and compression, location and topology independence, and energy constraints and controls. This evolution will challenge suppliers to rethink portfolio strategies, including sizing, pricing, integration, reference architectures and partnerships. This concept does not exclude mainframes or other systems tightly integrated as long as the resources can be functionally diverse, heterogeneous and location independent. Not all applications and workloads will necessarily mandate a fabric infrastructure of managed resource pools.

Figure 1. Impacts and Top Recommendations for Server Obsolescence

The server label relates back to the days of client/server computing in which a requestor, the client, sought the response of a server. The relationship was usually one-to-one or many-to-one (client[s] to server). Optimum responses were obtained by normally placing the client close to the server, thus leading to data terminals with a direct connection to servers or stand-alone host systems. However, the evolution to the Web, greater peer processing, the utility of the network using protocols such as TCP/IP and the higher speeds of network connections combined with virtualization are forcing a re-examination of the traditional computing architectural model to one based on any-to-any connectivity over wide-area networks. The network node represents a more modern concept of server. The node refers to a resource situated on a network at some location and in some form factor available to applications or workloads for execution. The resource is an extended concept beyond CPU that is inclusive of processor cores, memory, I/O interfaces, graphics and numeric processing elements, encryption processing and storage. Form factors now vary much more widely than the older use of minicomputers and mainframes. The hosts can comprise blades, racks, skinless, or other configurations comprising motherboards and associated processing, storage and I/O elements. Or the resources could be viewed as an aggregation of components represented by new terminology: multinode, physical or virtual, locally or remotely accessed within a common logical and managed construct.

Recommendations:

• Begin revising internal server road map plans from single function, separately configured systems to hybrids of storage, networks and servers.
• IT planners and architects must adjust procurement evaluations and suppliers to re-examine portfolios/SKUs as cloud and network-based virtual and distributed resources accelerate.
• Start to plan and prepare to reorganize the IT department's organizational structure and job roles; while traditional low-level administration and operational tasks will not completely disappear, plan to repurpose them to perform higher levels of management on shorter integration projects of fewer administrators and operators.
• Review the amount and type of integration projects and usage of system integrators required to put together server, storage, software and network systems and services.

Fabric-based infrastructures are playing an important role in the architecture of managed resources such as cloud computing¹. As the adoption of virtualization, clusters and internal clouds grows, fabrics will become a foundational technology of IT organizations, hosters and cloud services to create networks with the flexibility (e.g., rapid provisioning and load balancing) required to support highly virtualized server infrastructures where the VMs are no longer tied to physical servers. This will drive managed resources as a paradigm of integrated fabrics to 30% of servers shipped.

Servers will be only one part of the logical view of delivering services to the client. The number of ports, interfaces, cables, storage area networks (SANs), storage and system management will be inclusive in this logical view. As dependent variables subject to workload demand variance, location placement, interconnects across tiers, availability, distance to data retrieval and unique performance requirements (latency), servers will be subsumed by the converged infrastructures within the procurement process. Vendors may try to "submerge" component pricing to improve margins bundled within the larger fabric infrastructure. The greater the emphasis on system value, rather than component (commodity) value, the greater the opportunity vendors will have at creating uplift to margins. This systematized procurement will place added complexity and pressure on procurement specialists to unbundle the proposals and quotes to ascertain the true cost of the server, network and other components as if they were individual line items in the order.

Recommendations:

• Reconfigure cost models and BOMs of server acquisitions as part of resource pools composed of single and multinode network instances.
• When purchasing integrated systems, request at least two vendor bids and scrutinize comparative costs (beyond traditional methods of only comparing components), and also quantify and verify proposed operational productivity improvements.
• Request vendors of integrated systems to break out component costs in line items of server, storage, management and application software to determine logical integration premiums.

In the server-centric model, the roles of server, storage, application software and network administration have been separate, with different parts of IT organizations using specific toolsets for the tasks associated with their respective roles. Consequently, installing additional new servers or VMs often creates complexity as each management team serially signs off on different aspects of the server or VM configurations before it can be deployed. This process takes weeks to deploy a new application, server or VM, which slows the IT department's response to changing business needs. This friction-laden approach and methodology is increasingly unacceptable among global enterprises in highly competitive markets.

The outdated factor of the server concept is related directly to this management inefficiency. Therefore, IT organizations should think about the management organizational structure and process of server/storage/network management holistically, rather than as a fragmented piece-by-piece enablement issue. There are two choices: Continue with a segregated model where server administrators deal with the servers, network administrators deal with the network and storage administrators deal with the storage. Or, create a unified management structure as part of fabric-based infrastructure, where one set of administrators has overall responsibility for all configuration tasks in the fabric – whether servers, switches, or storage. As the latter becomes more prevalent, we do not predict the demise of the server but we do predict that the respective and constituent components of the resource delivery solution in the form of multinode clusters will become the new paradigm of architectural implementations.

As this confluence abets data center modernization, architects, planners, system analysts and administrators will be faced with new tasks, such as:

• How to assess performance responsiveness
• How to compare respective vendor solutions for functional and value-based pricing
• How to measure value over short-, medium- and long-term time frames, especially as greater functional diversity (e.g., OSs, cores, coprocessors, hypervisors, Flash, switches, solid-state drives [SSDs], etc.) and resource mobility occurs

• Make new calculations on life cycles, resource scalability, elasticity, VM mobility and placement optimization that emphasize the dynamic nature of the network versus the static generational improvements of the server.
• Weight intelligent automation (virtualization resource management, monitoring, self-diagnosing/correcting, node placement, added accelerators, etc.) across multinode networks (as the new competitive differentiator versus static hardware configurations).
• Create a metric (call it a "coefficient of friction") where the more fluid the parts blend together the higher the efficiency metric of the infrastructure (availability, performance, deployment, problem resolution); the greater the friction (e.g., manual interventions, slower reaction to business needs, slower life cycle maintenance, higher energy consumption and costlier upgrades), the lower the efficiency metric.
• Capacity planning and asset usage of the components within fabric and converged infrastructures must be closely and constantly monitored as they become subsumed into integrated systems; therefore, request detailed capacity planning and monitoring tools for these systems.

¹Examples of fabric-based infrastructures for managed resources include:

• Cisco Unified Computing System
• HP CloudSystem Matrix
• VCE Vblock
• Dell VIS Creator
• Egenera Converged Infrastructure
• Fujitsu DI Blocks
• IBM Dynamic Infrastructure
• NEC Dynamic IT Infrastructure