Network know-how

The aim of any implementation is to create, enhance or maintain a reliable network that falls within budget constraints. Unfortunately for the network manager, while budgets are often being squeezed downwards, the expectation is that the network should steadily become more robust and effective. NME investigates how network managers can best plan for and execute implementations.

  • E-Mail
By  Simon Duddy Published  January 23, 2005

|~|Rula-Ammouri-2_m.jpg|~|“We started studying in November 2003 and in May 2004 we went to United Business Machines with a request for proposal (RFP). We had looked at how we were connecting, how many users we had and what type they were, and also what kind of applications we were using. This gave us the information we needed to carry out a design and research the equipment we would need.” - Rula Ammuri, chief information officer (CIO), Jordan Telecom.|~|Implementing a network is never an easy task as there are a multitude of priorities to juggle, and at each step of the way it is imperative that the relationships between disparate parts work together seamlessly. On top of this, the right products must be bought to serve needs today and in the future and the project must fall within budget. There is no doubt that revamping the network is a daunting task whether you are overseeing a project, focused on an individual component such as design or working in a specialised area like wireless or security. This pressure cooker atmosphere has also had a positive impact, however. It has led to vendors and systems integrators being more focused than ever on demonstrating real value to customers. “Gone are the days of unlimited budgets and buying technology for technology’s sake,” says Yasser Helmy, business development manager at HP ProCurve Middle East. “Clients are now looking for a solution to address real business requirements in with constrained budgets,” he adds. There are many crucial elements that contribute towards a successful network implementation but the first, and sometimes-neglected step, is defining a company’s aims. It is only once these are clearly defined that decisions can be made on how the network should support these. Networks are definitely not ‘one size fits all’. What might be a good network for one company, won’t work for another. One of the most important areas to look at when defining business aims is availability. Although every company wants the network to be available all of the time, the cost of downtime varies greatly from business to business. Emirates Media is towards the higher end of this particular sliding scale and stipulated resiliency as one of the main criteria when it recently revamped its core network with Foundry FastIron 800 switches. “We put two FastIron 800s in each of the two main campuses and set up the network using virtual router redundancy protocol (VRRP),” says Derek Holland, head of Information Technology at Emirates Media. “This meant that if one of the core switches went down, the system would automatically compensate for the loss using the other switches, meaning there is high availability. With the company relying on advertising money for TV programmes we can’t afford downtime,” he adds. While it makes sense for Emirates Media to invest heavily in ensuring it has a highly available network, it isn’t that great a priority for many other businesses. For many smaller companies, cost is a much more important consideration. While there is no avoiding that a company will have to spend to succeed, the key aim is to spend wisely. One factor that helps to keep costs down this is being able to accurately predict the changing needs of the company. This is very difficult, given that the trajectory of a company is rarely a smooth one. Nevertheless, at each step the network has to be tailored to cater to business need. To prepare, the smart network and IT manager highlights scalable and flexible solutions that can see them through a variety of scenarios. When a company grows, ideally the IT department should grow with it. If new equipment needs to be bought, it is better that it integrates with and supplements existing kit rather than requiring an complete overhaul, which can have implications for downtime, as well as obviously costing more. “Selection of products should not only take into consideration current requirements and needs but also need to cater for any future changes,” says Yarob Sakhnini, regional technical manager, Foundry Networks. “Foundry’s product and design strategies are built on future proofing the networks of today,” he adds. This is true across all product groupings but is especially important with cabling, which often is designed with a 25 year lifetime. With the rapid pace of technology, it can be difficult to guarantee that a cabling system will prove adequate for 25 years. “Applications continue to demand more capability from the cabling infrastructure,” says Darren Stratton, sales director at Systimax Solutions, Middle East North Africa. “Over the last 15 years, speeds have increased from 1Mbytes/s up to 10Gbytes/s. That’s five orders of magnitude. In projecting future requirements, an organisation should count on the need for at least one more order of magnitude for every five years. As such, campus planners should know the types of applications that will be used at least through the next two generations,” he adds. Predicting which applications will be in use in five or ten years time is a nebulous task, many companies struggle to fully understand the applications they have now. For example and indeed many of the problems faced by companies on the wide area network are as a result of being surprised by the amount of traffic created by certain applications. Investigating applications therefore can prove a worthwhile exercise when preparing for a new implementation or when planning to increase the efficiency of the current one. When Jordan telecom upgraded its network from an aging ATM system with a Cisco-based Gigabit Ethernet infrastructure it thoroughly investigated its current situation before planning the upgrade. “We started studying in November 2003 and in May 2004 we went to United Business Machines with a request for proposal (RFP),” says Rula Ammuri, chief information officer (CIO), Jordan Telecom. “We had looked at how we were connecting, how many users we had and what type they were, and also what kind of applications we were using. This gave us the information we needed to carry out a design and research the equipment we would need,” she adds. United Arab Emirates-based systems integrator Emirates Computers has a long history of carrying out implementations in the region and the firm emphasises knowledge as a key step. “Understanding requirements and expectations are essential to the successful design of the network,” says Saeed Agha, networking unit manager at Emirates Computers. “It is necessary to understand traffic patterns, bandwidth requirements, server location and function, users communities, budget and future time frames for growth of bandwidth and applications. The major objectives are to ensure there are no bottlenecks, support future user growth and be able to cater for changes,” he adds. Knowledge is the key to the first stage of a network implementation. The company that does its homework and is aware of its business aims and applications, as well as having a well thought out projection of likely growth is half way towards a success. When it comes to putting the plan into action, companies follow a wide variety of routes, with some firms relying on an internal team, while other looks to consultants to a greater or lesser extent. Each approach can be valid but a thorough and well laid out plan of action is essential whichever approach is chosen. “Every step throughout the implementation has to be categorised with time scales and responsibilities. Failure to do that will result in a firefighting scenario later on, which is a network manager’s nightmare. So planning and more planning is required before actually doing the job,” says Sakhnini. Consultation can help a company to plug gaps and answer uncertainties. Large scale network implementations by their nature don’t pop up every day, and therefore experience may be lacking within the company. Among corporates in the Middle East, consultants are an often-used resource, whether they are independent specialists or advisers working within systems integrators or vendors. Dubai’s e-TQM College singled out consultation as one of the most important factors in the recent implementation of an Cisco IP telephony solution. “It was very important for us as our users were accustomed to a traditional PBX solution,” says Sameer Khoory, the IT manager for e-TQM College. “The new solution brought in features such as interactive voice response (IVR) and unified messaging (UM) and the company needed to know how to make a smooth migration,” he adds. If end user and consultant are well prepared, then the consultation period should go well. If the company has clearly defined its business aims, and has strong knowledge of its capabilities then what is needed in the new network should become apparent. While consultation is important, it can lead to over-reliance on consultants, particularly if the end user is dazzled by their experience and knowledge. This can lead to a loss in control and the end user accepting a solution that may not be suitable to the company. Rabih Itani, network and security manager at the American University of Beirut, says that it’s vital that the end user stays in control of the implementation. “The design can be reviewed, audited or even developed by consultants but it is important for any network manager to be on top of all the blueprint details,” he says. After the consultation period, it should be possible for the company to thrash out an appropriate budget and size up what solutions it can afford. The end user has to be aware of the soft costs as well as the more obvious upfront hardware, software and bandwidth costs. Items such as training, installation, configuration and ongoing administration can often prove more expensive across the lifespan of the implementation. Once the budget is in place, the company can go ahead and design the network, outlining the solutions needed to bring it into reality. There are many tools that can used to assist the network manager in this task, from making use of Microsoft Visio to sketch an outline of the network to utilising free vendor assistance. HP runs an online design centre from its German operation which provides free help on designing a network, albeit with a HP bias. D-Link has a similar design centre, which it is aimed at solutions providers, and guarantees a network design within 48 hours of submission of project details and drawings. It is at this time that the company must decide which topology suits its business best. While mesh topologies are currently in fashion, arguably the most popular configuration is a star. In terms of cabling, the most flexible topology is a hierarchical star in which a centrally located building acts as the hub with cables placed out to other satellite buildings. Connectivity between buildings is managed at the main cross connect in the hub building and new buildings can be added to the main cross connect. “A ring topology offers the highest reliability though,” says Systimax’s Stratton. “It achieves this by allowing the network to be reconfigured to bypass a failed node or broken cable. During a failure, networks designed with two counter-rotating rings can offer automatic restoration of the network by the active electronics but adding a new building to the network is more difficult within the ring topology,” he adds. After the bidding process, the design should be revisited with the selected system integrator or vendor and more details should be taken into consideration. It is a good idea to prepare an Acceptance Test Procedure (ATP) document that outlines how deliverables should be tested, which allows the end user to determine if the solution really works. Running the tests can be done via a pilot or after a full-blown implementation, depending on network conditions and requirements. Indeed piloting is not only an excellent way to test the validity of a prospective implementation but is also suitable for comparing competing solutions. The National Drilling Company (NDC) recently piloted three WAN optimisation solutions, which gave the company the opportunity to assess them in terms of administration, stability and speed, before it chose Peribit. If the end user has time, piloting can determine if the theory of the sales pitch translates into reality. Defining aims, planning and installing are all important steps in an implementation, but one often overlooked aspect of the job is what happens afterwards. It can be tempting for the network managers to rest easy but the maintenance of the network is critical. “One item that must absolutely not be overlooked is the administration and maintenance of the networks. The solution chosen must be simple enough for the user to administer, and not too costly to maintain,” says Saket Subramaniam, chief technology officer (CTO), Red Cube Systems. Post-installation testing and network stabilisation periods should help enable a seamless experience for users who are migrating from one network to another. Adequate support is also key in the post installation phase and indeed many local companies point to support as the most important quality they are looking for. No company wants to install an all-singing, all-dancing network, only to find itself on its own if the system hits snags down the line. “Our most important criterion was support. It was crucial that support was available on the ground, for example, if we wanted to re-patch the network frame,” says Khoory from e-TQM College. Despite its potential to make grown men cry, network staff need not lose too much sleep if they prepare well for implementations. One lesson network managers could learn is to place inter-personal relationships and teamwork on a high level of importance and not rely too much on technology to solve problems. “Passage of information and coordination between all parties from the end user through to the solutions providers is key,” says Richard Jasnau, divisional manager of telco operations & infrastructure at Sahm. “The hardware or software utilised within the network can be substituted or replaced but the information and personnel associated with the project ensure the success or failure of any implementation,” he adds. ||**||

Add a Comment

Your display name This field is mandatory

Your e-mail address This field is mandatory (Your e-mail address won't be published)

Security code