Matt Gebhardt
CS 522:
Computer Communications
Dr. Chow
Considerations
in Network Design and Management
Table of
Contents:
Abstract 3
Bandwidth Managers 3
Restoration and Disaster Recovery 4
High Availability Networks 4
Improve Performance 5
Network Management Systems 6
Network Security 7
Capacity Planning 8
Conclusion 9
Reference: 10
Network administrators have more tools and information at
their disposal now, than ever before.
Networks are changing every day.
Users are always demanding faster connections, more stability, reliability
and more features. Administrators need
to keep this in mind from the initial planning stages of the network to ensure
scalability, performance, and end user satisfaction. There are plenty of design concerns and this paper will address
some of them and attempt to provide a solid reference to those seeking solid
network design principles.
Bandwidth mangers can be
hardware or software based. After
implementing the hardware, the software gives the users the ability to break down
the traffic based on set criteria.
Similar to content switching, the users now has the ability to sort
network traffic by IP address, port number, or even custom parameters. Classification schemes vary from product to
product, the research in this field is advancing at a rapid pace and soon
optimization algorithms can become uniform.
Today bandwidth management is quite complex and only for experienced
network administrators. Handling TCP
traffic requires the examination of each packet while requiring each side of
the TCP connection to be fooled into thinking there is a long delay while the
bandwidth managers handles the traffic so they don’t retransmit the
packets.
Sometimes,
even the best laid plans falter and the network will go down. The problem can be simple or complex, but is
often difficult to diagnose. Some
network topologies conform to good design protocols and in the unfortunate even
of a cut cable or outage, can self heal.
Others are not quite as well designed, and the problems can bring
business to a halt. A wise business
accounts for these possible scenarios when designing the network to laying the
cable to future construction where a cable cut is possible. Business should have a detailed restoration
plan as well as the software to successfully restore and repair the
network. Additionally, network
administrators need to be ready to come in anytime to fix the problem. Network outages can mean lost productivity
or revenue for the business. Some of
which is time critical (i.e. Amazon.com around Christmas) and the revenue
cannot be replaced, damaging the firms reputation and possibly shifting
customers to a competitor.
While all networks are important, some are mission critical
and must be kept up at all times. 3Com
has set forth a good definition of a high availability network with the
following characteristics:
Resiliency — The ability of network and other devices to apply
preventive safeguards as a way of heading off degradation or failure.
Redundancy — The presence of multiple units of critical hardware
devices, such as network interface cards (NICs), switches, routers, and other
equipment and components.
Manageability — Network management is the manager's window into network
availability; it is essential for evaluating and tuning the network's
availability.
With this in mind, 3Com also prescribes some very intrusting
solutions. A Location-Based Strategy is their solution where the availability
of key network components is maintained at critical locations. The data center is a point for critical
business applications and should be upgraded with redundant systems to ensure
optimal performance and reliability.
The building or campus backbone also must be protected. These are high bandwidth links that connect
to everything from data centers to desktops.
If a backbone fails, most users will be unable to gain access to the
network. 3Com recommends the inherently resilient ATM protocol for backbone
technology. The WAN interface is another section that needs to be guarded. E-commerce relies heavily on the WAN
interface, and it is critical for building truly highly available
networks. Load balancing and link
redundancy can be used to alleviate the problem.
There are
many methods to improve network performance.
Load balancing, clustering and traffic management systems are a
few. These can be hardware and/or
software based, but will probably include a mix. Network administrators can even design customer applications that
work for their specific network. Intel is offering a new network processor with
should revolutionize the industry and significantly speed up some network. There have been methods to improve network
performance around for as long as networks themselves, but only recently has
the subject been researched in depth and solutions derived. New methods like LVS and content switching
will have a great impact on the industry and could prove to be a boon for any
company knowledgeable to take advantage of the opportunity.
Network
management is crucial to high available networking. Managing a network allows the administrator to trouble shoot,
optimize, or generate reports about the network performance. Some sophisticated network management tools
include:
·
Remote
Monitoring Software - installed on the switch allows information to be
monitored from a remote location
·
Management
Information Base (MIB) - agents that perform protocol analysis and determine
network usage
·
Service
Level Reports - Describe how the goals were met and provide invaluable clues on
how to increase performance in the future.
There are new tools coming out everyday, but most
are vendor proprietary and there are few standards. However, progress is being made, and these tools are allowing
administrators to “bulletproof” their networks and manage them more effectively
and efficiently.
Security is a major issue in today’s networking world. Information is key, and ensuring it arrives safely, accurately, on time, and to only the intended recipient(s) is of paramount importance. Sharing information across a many users and machines can leave a firm open to attack. There are many forms of attack and many ways to respond to a successful attack. Companies that are successfully compromised are not quick to share that fact with the authorities, the public, or the competition. If a firm’s system has been compromised, the customers may feel unsecure about future dealing with that firm, now perceived to be unsecure. Competitors will be quick to tout their security over the compromised site. The compromised site should be ready and willing to share information with the authorities to bring justice to the offender. This would also dissuade copycat attacks. The shared information would help more sites become secure. Sharing this vital security information with competitor would create a more secure Internet, increase customer confidence and lead to greater long-term profits. While short term profits may decrease in response to this radical idea, overall the industry will be more secure, customers more satisfied, and new unforeseen opportunities will arise. Network security is very important, and needs to be address at every stage of the network planning, design, implementation and maintenance.
There is never enough bandwidth. Users always want a faster and more stable connection. This is where strategic capacity planning comes in. When initially designing a network one must assume the basic infrastructure will be in use for a long time, there for it must be very scalable. One of the major costs and hassles is laying cable. One should understand this and realize that laying new lines will be very expensive, in terms of the actual cost and the opportunity cost lost because of the inconvenience. With this in mind, one must consider four factors when making capacity planning decisions:
User Dissatisfaction: Network loads are dynamic and without proper planning,
during peak hours, the increased load may force users to wait too long,
decreasing their satisfaction.
Productivity Decrease: Increased network traffic will mean a decrease to
productivity. Employees may have to
wait too long for their transaction to be complete and other work may
suffer. Alternately, they may think
there is an error, and send the query over and over again, further clogging the
system.
Budgetary Constraints: Funds need to be made available before the problem
occurs. This will ensure that it can be
scaled whenever necessary instead of fighting for funds from management.
Stability: Early Identification of problem areas and being proactive
can increase stability and user satisfaction.
Stability problems usually occur during peak loads when your systems are
most needed. Proper planning and
traffic monitoring are needed to fix this problem.
With the importance of networks increasing, it falls upon
network designers to ensure there networks will be able to deliver what their
clients are expecting. Using proper
design techniques, this can be achieved.
Ensuring the network is secure, scalable and meets the goals of the
business is achievable, but time consuming and expensive. However, the alternative is worse. Not thoroughly planning or sloppy design
will cause nothing but headaches for the administrators and users. Setting aside extra time and capital at the
outset should alleviate or significantly reduce these problems. Networks are important, and will continue to
be so for the foreseeable future, and ensuring proper design will allow the
business to achieve it’s goals and make everyone’s life a bit easier.
http://www.nwfusion.com/reviews/0406rev2.html
http://intranets.about.com/compute/intranets/cs/bandwidthmanagers/
http://www.ieng.com/univercd/cc/td/doc/cisintwk/ito_doc/snmp.htm
http://www.3com.com/technology/tech_news/highavail.html
http://compnetworking.about.com/compute/compnetworking/cs/qoscapacityplan/index.htm