Data center consolidation:

Takeaways from the pros Data center consolidation can be an effective way to cut costs by eliminating obvious waste such as redundant applications, unused files, and under-utilized hardware. However, cutting costs by way of consolidation requires more work than simply paring down servers and software. It requires planning and lots of it. In order to maximize the effectiveness of your data center consolidation, a host of issues must be considered:

• Changes in the physical space between servers and users
• Using or not using virtualization
• Flexibility to accommodate future changes
• Planning for possible technical issues
• Changes in local and market costs for labor and materials
• Changes in users' experience and expectations
• Security

In short, planning is the key. The articles here take a closer look at some issues with, and examples of, data center consolidation. They not only present information about data center consolidation from a rhetorical perspective, they also explore what some institutions have done well, what can be done better, and what shouldn't be done at all.

1. Overview

Enterprises are faced with a growing need for higher bandwidth in the horizontal and backbone to support such applications as Storage Area Networks (SAN), Network Attached Storage (NAS) and high performance computing. Corporate DCs provide Internet and other network services that are devoted to a single corporation. This typically supports:

• Intranet Services allowing communications within the corporate workforce
• Internet access for the workforce
• A Web Presence for customer support and remote access
• Hosting DCs are owned and operated by service companies that market web access, hosting, and other Internet based services. A wide range of these services can be defined:
• Collocation typically offers customers simple data center space with defined footprints (e.g. partial or full racks/cabinets, locked cages or private suites), network bandwidth, reliability, and security. Customers can install and manage servers, switches, firewalls, and web applications through on-premise visits and remote administrative tools.
• Managed Collocation provides extra on-site services such as:

• Equipment location and configurations for maintaining segregation of different hosted customers.
• Operating software and configurations for maintaining service isolation between different hosted customers. VLAN technology may be used to partition channel bandwidth and share server capacity.
• Additional private networks using VPN technology.
• Enhanced security to establish customer zones and prevent unauthorized or accidental access between different hosted customers. Locked cages or private rooms are typically used. Video surveillance and other personnel detection and sensory equipment are also used.

2. Standards

Cabling Standards for DCs have been drafted, these are the TIA/EIA 942 documents in the US and EN50173-5.200X in Europe. DC cabling generally follows many parts of the TIA/EIA 568B and IS11801 models for structured cabling, but these generally require adaptation to fit the high densities and sizing that are characteristic of DCs. Although TIA/EIA 568B and IS11801 do not specifically address data centers, many of the requirements can be applied. Overall, the bulk of cabling is comprised of the interconnection between the individual servers and the aggregation switches and from these switches to the backbone switches. A common cabling model is:

• Condensed horizontal and backbone cabling
• Including an IC (IDF) and an MC (MDF)
• Server links follow a horizontal zone cabling model terminating on consolidation points in server zones
• Servers within zones may be directly connected to the consolidation points

3. Data Center Cabling Selection

A cabling system may require a balance of both copper and fiber to cost effectively meet today's needs and support the high bandwidth applications of the future. Rapidly evolving applications and technologies are drastically increasing the speed and volume of traffic on data center networks. Ensuring that your cabling solution is designed to accommodate the higher transmission rates associated with these evolving bandwidth intensive applications is critical.

Due to the versatility and wide range of applications support, SYSTIMAX® Solutions has pioneered the development of UTP/fiber cabling systems to satisfy virtually all of a customers' data center network and building infrastructure needs. In addition to the performance of the cabling solution itself, the right cabling architecture needs to be chosen to optimize the investment and return for the particular building environment. Balancing cabling system cost versus the electronics, and also the ongoing management and flexibility of the solution is a key part of effective cabling infrastructure design.

The cabling architectures that are commonly used for DC structured cabling systems, require a mix of High-Density, High Reliability, High Performance, with Flexible Design Guidelines, Speed of Installation, Future-Ready and Easy to Use.

4. Copper or Fiber Solutions in the Data Center

- Which Is The "Right Choice"?
Network providers are often faced with the question of whether to install a UTP copper or multimode fiber cabling system to the servers. Unfortunately, there is not a clearly defined answer to this question. Most private networks require a mixture of both media to create the most cost-effective networks for voice and data across the horizontal and backbone segments of the network.

High performance Category 5e and 6 UTP, such as SYSTIMAX PowerSUM and GigaSPEED XL Solutions respectively, provides the lowest initial cost for today's Local Area Networks (LANs) up to rates of 1 Gb\s. Fiber-based networks can reduce recurring operational charges and clearly have higher performance, however with the introduction of Category 6A systems such as the GigaSPEED X10D Solution, the capabilities of UTP cabling have been substantially increased, pushing out the move to fiber for short distances.

Fortunately, SYSTIMAX Solutions provide complete solutions for both fiber and copper media. The choice of media depends largely on the customer's present and future applications and business situations. A selection made without considering these fundamentals has little chance of providing the best solution. That is why it is important to understand not only the capabilities of each media, but also specific customer needs. In order to determine which combination of twisted pair and/or fiber to install and in which architecture, each customer must evaluate their application needs, considering the various advantages of each cable type and their relative importance.

Cost, ease of installation, moves and arrangements, current and anticipated applications and the expected life of the system are typically major decision factors. Environmental considerations such as electrical noise and clean rooms may also influence the decision, as well as building type, industry sector and cabling system ownership. The anticipated need for low speed applications, short system lifetimes, and low initial cost might lead to a predominantly twisted pair cabling system. High-speed application, extended distances, harsh environmental conditions and graphics intensive multimedia applications might lead to a heavily fiber based system. Most systems will fall between these two extremes.

Given that there is some overlap in the customer base and capabilities for Category 5e, 6, and 6A cabling and multimode/singlemode fiber-optic cabling system infrastructures, an understanding of the customer's DC specific requirements is needed to recommend the optimal cabling solution. The decision will incorporate three phases -the definition of strategy, the design of the system, the cost effectiveness of the choice. For the definition of strategy, within budget restraints, each customer must consider and prioritize the following:

• The sophistication of their network applications
• The kind of traffic expected on the various portions of the DC network based on number of users, data transfer requirements of each user, LAN architecture, etc.
• The life expectancy of the network and cabling infrastructure
• The frequency of moves and changes
• The growth potential of their network over its expected life
• Any adverse physical conditions in the customer's data center

• Media considerations - performance, physical hazards
• Manufacturers support and warranty
• Cable containment including containment types (trunking, ducts, cable tray), and containment design (size, safety, segregation)
• Installation techniques and quality
• Adherence to standards
• Labeling, records and documentation
• Testing and certification
• Maintenance and services. When evaluating cost effectiveness, the customer should always think in terms of cost over the life of the cabling, rather than only the initial installation cost and also compare the cost to electronic hardware which will be replaced several times over the lifetime of the cabling. The lowest initial cost is not always the cheapest in the long run, however once the contract is placed it is difficult to change.

Choose the right system first time. Considering cost effectiveness should include the following:

• Initial installation cost: ensuring it covers adequately the specification to avoid unwanted extras and performance restrictions
• Administration: the network's ability to be easily and inexpensively reconfigured
• Maintenance: the effort required to keep the system operating
• Life cycle value: the assurance of a warranty covering the applications and hardware SYSTIMAX Solutions offer high performance copper and fiber cabling platforms that will cost effectively allow enterprises to implement DC applications and architectures. Customers need to be careful that in selecting a solution today, they do not limit themselves in the future. With advanced twisted pair and multimode fiber cable, connectors and apparatus, users can support all of their current applications, as well as their emerging and future applications.