Corporate Networks

1. Segment Your Shared Ethernet LAN

Are you looking for a competitive differentiator? Then look

internally. But look beyond IT. The information flow in your business process is what

holds the key. The smoother the flow, the better an organization is; the better is its

response time–a factor which can make or break the company’s fortunes. And the

answer to all this is the network infrastructure. Voice & Data presents seven

complementary technologies, which will beef up your connectivity infrastructure. And you

will gain the much elusive competitive edge. Read on:

1.

Segment Your Shared Ethernet LAN



alt="A segmentable hub utilizes the existing shared ethernet technology to provide a much more enhanced network."
align="right" hspace="4" vspace="4" width="300" height="186">Application time outs,

sluggish response times, and frustration are not new to network users. This is especially

true in India, where most of the networks are built around shared Ethernet (10 Mbps)

technology. Nowadays, it takes only a dozen of Pentium PCs running heavy applications to

reduce the Ethernet network to a crawl. Dual-speed Ethernet products (10/100 Mbps) are

being installed but a majority of users has a legacy of Ethernet LAN products to worry

about. And it is a catch-22 situation. You have to upgrade to Fast Ethernet (100 Mbps) but

at the same time you cannot simply throw away the legacy products. The picture is even

more serious considering the proclamation by several networking companies that the hub,

which is the main shared LAN product, is dead. You probably might be considering

approaching your bank for a major switch purchase. However, it is not all that bad. There

are companies who are putting enough research into developing value additions for the

shared LAN environment.

One such technique that has been developed and put into

practice is Ethernet segmentation. Segmentation is the process of dividing the

workstations in a network into workgroups or domains/segments. The easiest way to do this

is to put additional hubs and distance one workgroup from the another. This way there is

lesser number of collisions in each domain and, thus, the network is more responsive and

the bandwidth available per user is higher. What kind of value addition is that, you may

ask. However, in this basic segmentation process, there is one major hitch. The separate

domains will not be able to talk to each other at all unless you have a connecting device

(mostly a switch). And you are back to square one. You will have to invest in more hubs

and also a connecting device. This is where the segmentable hub (also known as the

switched hub) comes in. The segmentable hub is a single hub, which has two or more

internal segments. It may also have an in-built segment switch to take care of connecting

the various segments. It may also have a high-speed downlink for backbone and server

connection. Several of these segmented hubs can be stacked together with one of them

having the in-built switch and the downlink port. In this way the cost of adding a switch

is reduced while one has a fast reliable LAN, without having to change the existing shared

environment to a great extent. Add port switching to that and you eliminate the task of

manual assignment of segments to the workstations. The need for the administrator to go to

the wire closet and handle the physical cable every time a user moves to a different

segment is eliminated. Thus, what you have is an optimum LAN with equal division of loads

among different segments.

2.

Install a Remote System Alert



You have installed high-class networking technologies. But, will it help if you continue
using a dumb workstation? Most of the problem in a network happens at the desktop level.

Organizations often end up spending more in maintaining the desktop than buying it. What

if your desktops are more intelligent? Then lesser workforce would be required to identify

faults and rectify them. This intelligence can come in the form of monitoring or alerting

capabilities. This enables PCs to inform a management workstation when there is a problem

with the PC, such as a boot failure, unauthorized access, and over-heating.

The way to give this extra edge to your workstation is to

install a remote system alert in your network. There can be two types of remote system

alert, OS-present and OS-absent. Both types of alerts adhere to standards. OS-present

alerts can be generated using the DMI 2.0 standard, while OS-absent alerts follow Wired

for Management (WfM) 2.0 specifications. To install DMI on a desktop PC, one needs to have

DMI 2.0 component interface, a DMI management interface, and a DMI 2.0 service provider

program. In addition to these, a DMI-compliant management application like Microsoft SMS,

HPOpenView, and Intel LANDesk is needed to recognize and display the alerts generated by

the DMI components in the PC.

OS-absent alert systems has many more requirements, most of

them to do with NICs which is where the networking vendors enter. The requirements include

a PC equipped with DMI component interface which can be monitored independent of DMI and

the operating system, NICs with Peripheral Component Interconnect (PCI) specifications, an

interface between the PC and the NIC, and a NIC which is able to transmit an alert packet

onto the network whether or not the system is in a sleep state or in a pre-OS state. And

also a management application, which recognizes and displays alerts, and which is platform

independent. Platform Event Trap Format (PETF) specification is the system alert packet

format that a large number of management applications has started to support. vspace="4"

alt="A chassis product gives you the independence of adding to your network resources in batches, as your needs increase."> face="Arial" size="2"> name="3. Migrate from Standalone to Stackable or Chassis-based Products">3.

Migrate from Standalone to Stackable or Chassis-based Products



If your employees download and exchange heavy files internally and from the Web regularly,
say no to hubs. If they are not clamouring for bandwidth today, they will definitely in

the future. And a hub will not suffice your needs very soon. A hub is based on the sharing

of bandwidth, and if there is a large number of desktops actively connecting to it for

bandwidth-extensive activities, the chances are that some of the users will hog the

bandwidth available, leaving the others waiting for their files to open up.

Yes, there are number of networks who still do not require

large amount of bandwidth or who do not have too many desktops in the network. Cases in

which a Fast Ethernet hub allocates better speeds than say an Ethernet switch. There are

very few users or at the peak usage only a few users are using the network. That is

probably because you still have not provided the inevitable Internet connection to your

individual workstations. You can be happy till that moment. But for how long?

It is always better to plan ahead. Scalable solutions are

the need of the hour. The stackable hub and switch is a good way to grow your network with

the growth of your organization. Apart from saving on unnecessary costs, stackable hubs

and switches also have advantages in the architecture of the network. Instead of having

separate IP address for each hub, a stack will require just one IP address. Similarly you

will lessen the cost and managing of several individual cables for separate devices. The

stack will require just one cable connecting to the core device.

In the case of the core switch, the option of modularity is

usually offered through a chassis-based switch. Scalability is through a switch fabric

that allows you to expand switched network as and when required. Each chassis has several

slots to fit in additional modules as you grow. The modules could be of Ethernet, Fast

Ethernet, Gibabit Ethernet, Token Ring, FDDI, ATM, Remote Access Server (RAS), cable

access, DSL, and so on. This way, you keep your options open for future. If some of the

cutting edge technologies become tomorrow’s standards, you have a module that can

take care of working with that technology.

4.

Get Your Switch to Route for You



alt="Switch routers deliver all the routing functions of a traditional router, but at a much higher speed."
align="right" hspace="4" vspace="4">When it came to the core of your network, routers

always used to be in the mind. Not anymore! Today, switches have taken on routing in a

major way. Traditional routers, which once formed the core of enterprise networks and

still dominates the mother of networks called the Internet, is seriously threatened as far

as its performance is concerned. Several enterprise users migrating to a next-generation

network are not anymore considering the router as the first choice of core networking

device. The feeling is that traditional routers are no match for the high bandwidth

requirements on today’s networks. High-performance servers, corporate intranets,

groupware, and multimedia and mission-critical applications have all contributed to

network congestion. Network traffic is also very unpredictable with end users getting

highly mobile and Internet-centric. While software-based routers have rich features, they

cannot simply cope up with this kind of demand. At today’s rate, millions of packets

have to be pumped through the network per second. And traditional routers, simply, are not

designed to forward packets with such gusto. At the same time, though fast, Layer 2

switches lack the functionality to be installed in your backbone. So, what you want is a

product that has the performance of a switch while having the same level of functionality

that a router has. Now we have the switch router. Both Layer 3 and Layer 4

switch are options for your routing purpose. A Layer 3 switch is a combination of a switch

and a router. It is like a switch, which can determine forwarding paths, checksum header

validity, verify packet expiration, automatically add statistics, and secure the

transaction–all routing functions. Since it is able to operate with information at

Layer 3 of the OSI model, a layer in which routers operate that it is called a Layer 3

switch. However, the routing ability is somewhat different in the case of a switch router.

While routing in a traditional router is based on software, a switch router is ASIC-based.

This has resulted in the switch router’s ability to forward packets much faster than

a traditional router. Now Layer 4 switch makes further value additions to this, by

inculcating information pertaining to the Layer 4, like TCP/UDP port information and

application-level controls. One of the advantage over Layer 3 switch, for instance, is

that prioritization of traffic by type of applications is possible. While Layer 3 switches

had little usage in the WAN atmosphere, the Layer 4 switch has even started threatening

the router in the area it considers its domain. To be scalable upto the service provider

networks, the Layer 4 switches today takes on massive table capacity, multi-gigabit

non-blocking backplane, low latency, and seamless scaling.

In the first place, switch routing reduces the number of

devices to be managed. It also comes out cost-effective in comparison to having both a

switch and a router. More importantly the networks of tomorrow will have be built around

the demand that is foreseen. And several technological changes have already altered the

demand. For example today much of our work is based upon communication–the Internet

and e-mail are extensively used in the work environment. The trend today is that no matter

how much bandwidth is allocated, the user is always thirsty for more. You can say, how

far, how much more? The answer is not simple. But you better have it when it comes to that

point.

5.

Start Insisting on CIR and Flexible Bandwidths



A large number of companies which are national in nature or have offices in multiple
cities have either gone in for connecting the remote company locations with either VSATs,

leased lines, or through an EDI service provider. But chances are that the connectivity

link provided is just not good enough to run the applications that you want to implement

across your company. The wide area connectivity has always been and is still the main

stumbling block when it comes to migrating to the next-generation networking

infrastructure. Ability to provide flexible bandwidth and Committed Information Rate (CIR)

are a must for service providers in the US and Europe. In India, where a leased line often

takes days to be up after being down, these terms sound a dream. The incumbent operator

did not bother about quality of service at all. A reason why many corporates switched over

to VSAT service providers, was the perception of the latter as being more professional.

If you have leased a 64 kbps link, what is the guarantee

that you will not need much more capacity during one of those bursty peak hours when you

are, for instance, videoconferencing with a colleague in another city? Also, do you really

need 64 kbps of bandwidth throughout the day. There are times when your employees just

exchange files and e-mails which require much lower throughput. In such scenarios, we

often would like to specify the deliverable bandwidth in a much more customized manner.

Broadband access technologies like managed TDM, Frame Relay, cable, and DSL can deliver

this need. In India, these services are limited but not absent. Companies like IBM, TMI,

and GlobalOne provides such service, though they may not be able to deliver the same kind

of quality of service that companies in the US are familiar with, because they have

virtually no contol over the domestic leg of their networks. There are also a number of

ISPs and telcos planning to provide these services to you in the future. But, almost all

of them have not made the investments on their infrastructure to be able to provide these

services. One exception is Satyam Infoway which recently announced its "Frame Relay

on ATM" services. CIR is an important thing that the user has to bring to the

notice of the teleco. The user should define the speed that he requires and that

throughput has to be delivered everytime he uses the network. A usage pattern report has

to be submitted to the user as proof that he is getting the bandwidth that was agreed

upon. This practice has not been adhered to in the past. This was because the networks of

the monopoly DoT just were not good enough to provide any throughput guarantees. And there

were no other service providers as an option. Even today, as most of the service providers

still depend on the DoT for the network backbone, guaranteed bandwidth are still far away

in spite of the entry of private operators.

Users have very little options other than to bear with the

monopoly operator’s "take it or leave it" attitude. It is not that the

telecom service provider cannot upgrade his network to provide better services and also

enable others depending on his network to provide the same. The fact is that such value

additions are just not his priority. But, if you as a user do not make noises, it is

likely that the operators will never care. And you will always be cheated in paying huge

costs for a link that deserves much less.

6.

Telecommuting is the Way



alt="A Virtual Private Network (VPN) can be set up by installing remote access products and tying up with a carrier."
hspace="4" vspace="4" align="right">As the competition among corporates gets intense,

employees cannot be out of touch no matter where they are. If there is a query from a

client in whose site a travelling software engineer has done some work, the engineer has

to be traced out wherever he is. And he should be able to communicate with his teammates

back in the office. Likewise, a remotely located salesman needs to check back with the

goods status available on the company’s central server, to be able to provide a

client with a despatch commitment. In such situations, a remote access solution is needed

to enable the employees to access the company’s web site/server. Then, there would be

many employees in your organization who would also like to use some of their time in their

home to check and update their jobs. An increasing number of CEOs have attested to the

fact that their productivity improves with the ability to telecommute.

Installing a RAS and implementing a Virtual Private Network

(VPN) are some of the ways to enable telecommuting and remote accessing. By installing a

RAS in your organization, you are allowing your employees to log in to your resources by

dialling into the port in the RAS. Here, local users are the ones who will benefit the

most. This definitely allows your CEO to access the resources in the office sitting in his

home. Though this allows your employee to dial in from anywhere, the prohibitive costs

associated with calling in from a different city or a different country inhibits the

employee from doing so.

This dilemma is where the VPN perfectly fits in. VPN allows

for remote access to the branch offices, residences, and mobile employees even if they are

situated in another continent. It was the advent of the Internet, which pushed the

development of VPN. The Internet’s local points of presences allow users to access a

shared data in the form of a web site by just making a cheap local call. The technology

for accesing web content and e-mails has now been expanded to accessing LAN resources to

address the needs of the business user. VPNs establish a dedicated and secure path in a

shared public data network to ease the transmission of data between the local POP and the

corporate network. The dedicated and secure path/tunnel is established between a network

access point and the destination point where the path terminates. The network access point

has equipment, which encapsulate the packets so that they can be sent over the secured

tunnel. For this purpose, the protocol used are Point-to-Point Tunnelling Protocol (PPTP),

Layer 2 Forwarding (L2F), and now Layer 2 Tunnelling Protocol (L2TP) which is specially

designed for having a secured VPN over the Internet. 7.

Structure Your Cabling Infrastructure



If you are still sourcing your cables from a particular vendor and the cabling components
from another, then put an end to this practice. Mixing and matching your cable may prove

cheaper, but will the infrastructure be good enough for the fast and powerful active

components like hubs, switches, and routers that you will install on that? Several

research studies reveal the fact that more than 50 percent of the network problems are due

to cable faults. Unless you have a fast and capable cabling infrastrustructure, your

network will never be able to measure up to the speeds and performance that you expect

from it. Mix and match only adds to the headache of rectifying cable faults, which will

happen more often unless an expert implemetation has gone into the infrastructure.

The practice that is getting hotter day by day is the

concept of structured cabling. The basic difference of this, apart from others, is that

structured cabling is a solution. It is not just products as in the case of the mix and

match approach. The first and foremost aspect that is decided before going in for a

structured cabling solution is the roadmap that a corporate wants to take in terms of his

network requirement. Cabling is a long-term investment. Your system should have a large

number of outlets for nodes planned out in the first place, or at least the architecture

should have provisions to add up several more in the future. Cabling is best done along

side the construction of the building, which is to house the servers and desktops. If the

planning is not done properly, the whole cabling of the network might have to be done once

again to the extent that one may have to dig the ducts out or walls have to broken to

extend the connection and add more outlets. This is not only tedious and costly, it breaks

the work routine of an organization.

It is best to outsource the job of implementing the cabling

infrastructure to a company that specializes in cabling activities. Large structured

cabling vendors have a number of channels that have specialized in installing the

structured cabling solutions that they offer. Companies like Lucent and AMP have groomed

many a company to do the implementation. They have in fact training programmes in which

employees from leading SIs and NIs are educated about the cabling technologies and their

products. After training, these persons are certified by the vendor.

Once a cabling project starts, the specialists have in

their hand the corporate’s outlook and needs. Based on several factors, the outlay of

the network infrastructure is worked out. The media of cables is based upon the location

of the network. If the network has to be very wide with sub groups to be connected over a

distance of more than a hundred metres, a fibre cable could be the choice. If the network

is to be in a few floors of a building, the choice could be copper. Then, the question is

which type of transmission technology and speeds to go in for, STP or UTP: CAT5, enhanced

CAT5 or CAT6? All such aspects are looked into by the integrator/installer and then

brought to your notice. Together a decision is taken and the infrastructure is

implemented. The end solution is a much better one because the installer has understood

your needs, your future plans, and the chances are that he has installed similar

infrastructure before. A good foresight is needed to choose the right kind of

cabling at the right price. For example, if your company is to remain a small unit for

times to come, where is the need for a Gigabit cabling infrastructure. Chances are you

will never need that amount of speed and bandwidth for the lifetime of that infrastructure

(15 years approximately). The opposite is true; if you have a floor with 50 employees

presently but your company will have more than 500 employees occupying five more floors

above that one in the same building, is it enough to have install a CAT5 UTP system? You

probably might have to upgrade to CAT6 in another three years. A lot of the choice depends

on the type of applications that your company will go in for. If voice integration and

applications like browsing, videoconferencing are part of your daily routine, your company

definitely needs to consider a fibre backbone, unless in the future you will need to

totally rehaul the thick co-ax or UTP cable that you have presently. It always saves a lot

of money and bother to be proactive. In these times, it has also become a matter of

survival.