1. The network “cloud” shown in configuration diagrams is typically thought to be highly
specialized and the domain of telecom providers. Understanding the components within
the cloud enables organizations to control costs and optimize network performance
by matching specific applications to appropriate technologies and architectures. This
white paper will arm you with an understanding of these components—providing a key
advantage in working with technology vendors and telecom providers to develop business
continuity and disaster recovery solutions.
Networking the World’s Business Data®
Demystifying the Network Cloud
2. 22
Common depictions of networking equipment and configurations are generally more detailed
at the ends where whatever on-site equipment the client uses to create, transmit, and store
information is located. Between these end user locations is usually an irregularly shaped
cloud diagram with names like WAN, MAN, IP Network, ATM, Frame Relay, T3/E3, etc.
The cloud is generally considered to be the province of the telecommunications provider. A
typical network cloud includes the following components:
Telecom offices: the central office(s) of the telephone company building, where their wire
runs and where they house the telephone equipment such as switches, routers, and cross-
connects. This equipment, as well as the office itself, is hidden within the cloud.
Cable route: the path – a street or overhead pole line or even a microwave path – that
connects the telecom office to the customer site, or connects a number of same-company
telecom offices to each other.
Regeneration points: used for long-haul fiber optics and copper, these points boost the
signal that physics would otherwise degenerate after a certain distance.
Carrier meet-me rooms: the neutral location where different telephone companies
interconnect with each other.
Copper or fiber cabling: the physical cables that connect everything.
As you can see, the components of a network cloud are similar to the components in the rest
of a network diagram – it includes a physical space, cables, and hardware – all focused on
moving bits from one place to another. The great computer “secret” is that everything is just
moving 0s and 1s; the same is true in the network cloud.
Connectivity protocols and architecture
As discussed, your network cloud may include copper pairs, fiber optic cabling, and the use
of other transmission media including point-to-point radio (microwave), point to multipoint
radio (LMDS, MMDS, WiFi, etc.), and even laser driven lightwaves shot through the air. These
communications will be managed by a variety of protocols including Ethernet, TCP/IP, ATM,
Frame Relay, SONET, SDH, Private Line, MPLS, ESCON, and others. The data transmissions
will run at a variety of speeds ranging from DS-1 (also known as T-1) which is made up of 24
DS-0 channels for an overall speed of 1.54 Mbs, E-1 (the European equivalent of DS-1 which
includes 30 DS-0s), DS-3, E3, OC3, OC12, and Gigabit Ethernet.
The choice of the transmission media and managing protocol is dependent upon any number
of factors including distance between locations, cost, physical facility availability, needs for
privacy and security, and the requirements of the specific application. While engineers may
enjoy debating the merits of any particular methodology, the simple truth is that one is not
inherently better than another. Protocols work differently depending on the equipment, and
each application has a most appropriate speed.
Table of Contents
Connectivity Protocols
and Architecture 2
Physical Components 4
Riser System 4
Lateral Connection 5
Local Loop 5
Cross Connect 5
Long Haul Bandwidth 6
Summary 7
3. 3
The network cloud may also include the services and capabilities of several underlying
telecommunications providers linked together at central offices, meet-me-rooms, distributing
frames, building basements, and under the streets in manholes and cable vaults. Network
architectures can be a tree and branch configuration, with individual circuits homing back to
a central point, a mesh with multiple interlocking points, or a ring with the capability to move
data in either direction and immediately reconfigure in the event of a failure. Once again,
there are not necessarily right or wrong network architectures, but rather choices are made
depending on the demands of a particular application in regards to time sensitivity and data
latency.
All of these requirements, which affect price, performance, security, and reliability, need
to be taken into account when matching the application to the technology. Also keep in
mind that disparate applications and technologies can be complementary and this creates
opportunities to optimize connections and reduce overall costs.
3
The Network Cloud
The network cloud in a configuration
diagram includes:
• Telecom offices – housing wires,
switches, routers, cross-connects, and
other hardware
• Cable routes – path connecting sites
• Regeneration points – signal boosters
• Carrier meet-me rooms – interconnecting
carriers
• Cabling – physical copper or fiber cable
connections
Telecom offices
Cable routes
Regeneration points
Carrier meet-me rooms
Cabling
4. 4
Physical components
The network cloud may also include any number of connections and management devices
including routers, switches, bridges, multiplexers, de-multiplexers, regenerators, and cross-
connects.
Riser System
The riser system is the in-building wiring and cabling that links the various floors and suites
with the telecommunication vendor’s Network Point of Presence or NETPoP. The riser system
generally terminates in a distributing frame where the telecom’s outside wiring meets the
building cable.
This outside wiring is either twisted copper pairs in a bundled cable (which can range from
anywhere from 25 pairs up to hundreds of pairs) or is fiber optic capacity terminating in
electronics. In some cases both copper and fiber terminate in the building. Depending on the
building, there may be incoming facilities from either a single or multiple vendors.
Because telecom companies sell capacity on their cable plant to their competitors, multiple
companies serving a customer will not necessarily mean that more than a single provider’s
facilities terminate in the building. This is a very important consideration when planning route
diversity in a disaster recovery situation.
The Network Cloud
The network cloud includes a number of
points of connection and interconnection.
Lateral Connection
Cross Connect to LAN/WAN
Local Loop
Local Loop
Cross Connect
A
B
C
D
E
F G
Lateral Connection
Cross Connect to LAN/WAN
Local Loop
Local Loop
Cross Connect
A
B
C
D
E
Riser
System
Riser
System
5. 5
Lateral Connection
Connecting the building to the cables running under the street is the lateral connection. The
lateral is a spliced cable connected to the main cable, generally at a manhole. The important
takeaway about the lateral is that it is expensive to construct and, particularly in the case of
fiber optics, not every building is connected to the main cable via a lateral.
When choosing a carrier, it is important to ascertain not only whether the provider’s cable is in
proximity to your location, but also whether the necessary permissions and construction have
been completed to connect your building to the cable, which otherwise might just be “passing
by” the building. Once the lateral is put into the building, it is necessary to connect it to the
building distribution frame so as to have access to the cables in the building risers.
Local Loop
From the building distribution frame, through the lateral, and along the cable back to the
provider’s central office is the portion of the outside plant commonly known as the local loop.
This terminology comes from the common twisted copper pair scenario where the two-wire
connection back to the central office effectively does provide a loop.
It should be noted that in the case of DS-1 and DS-3 service, the loop is actually four wires,
with one pair for sending and a second for receiving, hence private lines are sometimes
referred to as four-wire circuits.
This loop architecture means that in most cases, and particularly those involving private line
services, data transmissions will leave the customer’s location, travel to the central office,
and then travel out across a different path to the end destination. The positioning of the
central office in the middle of the circuit is an important consideration. When planning data
transmissions, there is always a central office in the middle of a circuit that links buildings,
even when they are in close proximity.
Cross Connect
The central office is a point in the provider’s network where cables from an area converge at a
single central point. There are often 3 or more main cable routes branching out from the central
office to serve customer locations in the wire center area.
Within the central office is switching equipment for routing voice and some data services as
well as cross-connect points to enable cables from other routes radiating out from the office to
be connected together.
Both ATM and Frame Relay switches may be located in a central office and the connection from
the switch back to the customer location, while often referred to as a Frame or ATM connection
is in reality a high-speed private line. The central offices associated with the ILEC (incumbent
local exchange carrier, a.k.a. the “Bell” companies) generally link to another regional office, or
tandem location, whose function is to aggregate traffic from multiple local central offices.
6. 6
As a result of central office “disasters,” including the Hinsdale fire at Illinois Bell in 1988,
local central offices now generally have diverse routing, including ring architectures, to ensure
that the loss of a single office or connection link does not adversely impact communications
across an entire region.
Long Haul Bandwidth
When referring to central offices, telecoms often define them by number. A Class 5 central
office is the one closest to the customer. A Class 4 central office is a tandem location that
aggregates and switches traffic between the subtending offices and also links to other
tandem locations. In the interests of cost savings, competitive carriers generally link to the
incumbent’s network at the Class 4 offices, although economics and customer concentrations
may also dictate connection at certain Class 5 locations.
Other carriers’ offices are often found in space shared between multiple carriers with access
to long haul fiber routes as well as the capability to interconnect to the local ILEC of primary
importance. These buildings are referred to as Carrier Hotels, collocation centers, and meet-
me rooms. The important consideration of these facilities is that they contain fiber optic cable
terminations and the ability to connect to other carriers’ networks to route and exchange
traffic.
The key point here is that carriers tend to collocate their facilities since none of their networks
go everywhere, making it necessary to link with competitors in order to complete linkages to
points where their own facilities do not go. In the case of the local telephone companies, this
generally means they must connect to a long distance carrier such as AT&T, Worldcom, Level3,
or WilTel. For these long distance companies it means that they need to link to the facilities
of a company providing local service. While it is true that the Federal Communications
Commission is slowly letting the former Bell companies into the long distance market, it
is also true that, with the exception of Qwest, these local RBOCs (Verizon, BellSouth, and
SBC) do not have long distance facilities of their own. Accordingly, they still must connect to
another carrier when it is necessary to go out of their serving territory.
7. 7
Summary
As you can see, the network cloud includes multiple facilities and multiple types of circuits
connecting to and from several providers. These provider networks come together at several
common points, creating the opportunity to use more than one provider to meet your needs.
Grooming the network
Data connections are generally purchased on an application-by-application basis, generally
under long-term contracts ranging from 12-36 months. These connections run at different
speeds and may encompass several types of traffic including both voice and data.
With multiple terminations in particular buildings and between common points on the
networks, the carriers will “groom” their network to group facilities onto a common platform
and reduce their overall costs of service. While carriers constantly review their internal
networks for grooming opportunities, this is not a service they generally make available to
their customers.
Grooming involves reviewing all the components of a network, grouped perhaps by building
or by customer. This activity requires an inventory of all circuits which then can be mapped
between locations, looking for opportunities to physically combine the circuits. Grooming may
include changing the types of connections, or combining a number of smaller connections
onto a larger line to reduce management overhead. The goal of network grooming is to
improve the overall efficiency of the network and reduce costs.
This is one reason why some buildings have fiber connectivity and others do not. It simply
became economically advantageous to replace expensive copper facilities with a more
efficient fiber connection, reducing overhead and creating the opportunity for profitability.
Likewise, the closer to the customer location those services are multiplexed together, the
more efficient the network will be.
Optimizing your cloud
While the network “cloud” can remain an amorphous blob on your diagrams, a simple
understanding of its components and the ability to optimize the network connectivity within
the cloud can lead to savings opportunities.
For help with improving your price performance, look for a company that understands the
piece-parts in the cloud, and also understands what the cloud can do to meet your storage
networking needs.
