ious versions at the Internet Archive. Yahoo! is not affiliated with the authors of this page or responsible for its content.
Rapidly responding to data center changes with structured cabling
Rapidly responding to data center
changes with structured cabling
By Mark Lewis, IBM Networking Services;
Christian Dicht, IBM Networking Services;
and Thomas Oakley, IBM Information Development
IBM Global Services
September 2004 Rapidly responding to data center changes with structured cabling
Page 2
Rapidly responding to data center changes with structured cabling
Page 3
Introduction
Over the past 10 years, the data center environment has expanded beyond the
traditional glass house (mainframes and attached storage devices) to include
distributed server networks and storage area networks (SANs). To increase the
performance and availability of these data management networks, companies
are migrating the overall system configuration from the traditional, dedicated
server-to-storage direct connection to a switched server-to-storage shared
connection using fiber optic cabling. The migration to switched connections
enhances overall system availability and data throughput, but, at the same
time, greatly increases the complexity of the fiber optic cabling.
Because of this increase in connection sophistication, the need for planning
fiber optic cabling systems has never been greater. A properly designed fiber
optic cabling system is an essential component that can help you manage
system complexity and facilitate future growth and reconfiguration.
This white paper compares two generic methods of implementing fiber optic
cabling in the enterprise data center that are used today:
Direct point-to-point connections between system equipment ports, using discrete
fiber optic (two-fiber) jumper cables
A structured fiber optic cabling system, using a centralized connection area
known as a central patching location (CPL)
The comparison looks at how each method affects three critical factors in
managing overall system availability:
Reducing the amount of time required to plan system changes that include fiber
optic cabling changes
Reducing the actual system downtime required for system upgrades and changes
Reducing the amount of time required to recover from an unsuccessful system change
2 Introduction
3 Direct point-to-point connections
6 Structured fiber optic cabling
system
7 Managing change
11 Benefits of structured

fiber optic cabling

Return on investment

Scalability

Availability
14 Summary
14 For more information
Contents Rapidly responding to data center changes with structured cabling
Page 2
Rapidly responding to data center changes with structured cabling
Page 3
Direct point-to-point connections
The point-to-point connection method uses discrete fiber optic (two-fiber)
jumper cables that run directly between the active system equipment. Each
fiber optic jumper cable connects a single port of one device to a single port of
another device. This is often the default method for data center cabling because
it requires little up-front investment or planning. Additional fiber optic jumper
cables are added as needed.
A. Jumper cables run directly
from device ports in the servers
to device ports in the storage.
B. As switches are added to
increase the overall system
connectivity, additional jumper
cables must be connected to all
the devices to provide sufficient
physical connectivity.
Servers
Storage
Servers
Switches
Storage
A. Direct point-to-point
B. Switched point-to-point
Figure 1. Growing complexity of point-to-point connections
Figure 1 illustrates how cabling with discrete fiber optic jumper cables starts
simply, but the complexity quickly grows as switches are added into the system
hardware configuration. Rapidly responding to data center changes with structured cabling
Page 4
Rapidly responding to data center changes with structured cabling
Page 5
The direct point-to-point connections seem, at first glance, to be the simplest
and most cost-effective method. As you add new equipment to the system,
you simply run additional fiber optic jumper cables under the raised floor to
connect the new equipment to the existing system. Indeed, fiber optic jumper
cables have become a commodity in todays data center environment with
many suppliers offering them, even over the Internet. However, as both Figure 1
and Figure 2 illustrate, the simplest direct point-to-point connections become
complicated as a system grows and switches are added. Fiber optic jumper
cables are installed under the floor and fan out to connect servers to switches
and switches to storage devices, without an organized management approach.
Soon it is difficult to determine the origin or termination points of any of the
fiber optic jumper cables.
Figure 2. Actual fiber optic jumper cable installations Rapidly responding to data center changes with structured cabling
Page 4
Rapidly responding to data center changes with structured cabling
Page 5
The usual method of installing fiber optic jumper cables is also a result of the
uneven growth that many data centers experience. The uneven growth results
from unexpected user demands and system upgrades. Hardware technology
and software applications continue to change, providing IT managers with
difficult choices every day.
In a global marketplace, internal and external forces impact the decision-
making process. You are constantly confronted with the following questions:
Can you add new applications and new users, perform system upgrades and still
reduce complexity?
Can you perform a system upgrade to facilitate growth, while efficiently managing
the change to the fiber optic cabling infrastructure?
Can you keep pace with your competitors and respond rapidly to change by reducing
the planned outage window?
This fast-paced, often reactive, environment can lead to a lack of long-term
planning, especially when it comes to fiber optic cabling. It is this lack of plan-
ning, without a systematic approach, which leads to tangles of orange (and now
yellow) discrete fiber optic jumper cables under the raised floor. The longer the
cable chaos is allowed to exist, the more difficult it becomes to rectify. With
this type of fiber optic cabling, if a system upgrade yields unsatisfactory results,
will it take minutes to restore the pre-change system configuration, or will it
take hours?
How can cable chaos be avoided? How can fiber optic cabling be planned
and managed so that it actually helps you manage the system? The answer:
use a systematic approach to fiber optic cabling that includes documented cable
pathways, designated cabling connection change areas and a methodology
that enables most of the change activity to occur independent of the planned
system outage window. Rapidly responding to data center changes with structured cabling
Page 6
Rapidly responding to data center changes with structured cabling
Page 7
Structured fiber optic cabling system
The second method of implementing data center fiber optic cabling is using a
structured cabling system. This method uses multifiber trunk cables and fiber
optic connection panels (known as
patch panels) to establish defined cable
pathways and designated cabling connection change areas in the physical
cabling design. As soon as these known cable pathways and cabling connection
change areas are established, the fiber optic cabling for system hardware
modifications can be isolated to the designated change areas in the cabling
infrastructure. This greatly simplifies cabling modifications by managing the
change activity in the defined locations rather than disconnecting, moving and
reconnecting discrete fiber optic jumper cables scattered under the raised floor.
The structured fiber optic cabling system approach, illustrated in Figure 3,
shows how the system approach connects all the device ports to the main or
CPL. This approach relocates the equipment ports away from the active,
running equipment and creates a designated cabling connection change area.
The actual server-to-storage connections are made at the CPL, using short,
manageable fiber optic jumper cables that are not placed under the raised floor.
A. Server-to-storage connections
B. Switched server-to-storage connections
Servers
Switches
Storage
Storage
Servers
Zone
patching
location
(ZPL)
CPL
A. Servers and storage devices
are connected to the patch panels
located in the CPL. This creates
the defined connection change
area, the CPL, and isolates the
active equipment from future fiber
optic cabling changes.
B. As additional equipment
is added to the system
configuration, it is first connected
to the CPL, and then connected
to the active system by making
the necessary fiber optic jumper
cable connections at the CPL.
Figure 3. Structured fiber optic cabling
CPL Rapidly responding to data center changes with structured cabling
Page 6
Rapidly responding to data center changes with structured cabling
Page 7
When switche