Building the bullet-proof network

Today, high network availability remains the best defence against the dire consequences of downtime.

Gary Boniface, CSO Manager at 3Com SA, discusses how coupling hardware - that allows for a redundant network design - with effective data storage and network management strategies can enable companies to maximise uptime and greatly decrease the odds of a catastrophic network outage.

From banking to government, enterprises today rely on their information infrastructures to exchange more data than ever before.

Whether utilising customer databases to analyse promising cross-selling opportunities, or converging voice and data communications over a single architecture, companies are increasingly turning to their networks to cut costs, improve services, and fulfil their most critical business objectives.

However, the more a business depends on its network, the more devastating a major failure can be.

According to some research companies, a single hour of lost connectivity, for example, can cost an airline reservations desk more than R85 000, an order fulfilment centre almost R200 000, and a retail brokerage firm a staggering R6.3 million.

Also, a recent study from a leading university says, less than one in 10 companies survive a catastrophic data loss. The rest either close within two years or never reopen at all.

When the network supports every aspect of communications - data, video, and voice over IP - traffic must move as quickly and smoothly as possible with minimal packet loss.

Using Gigabit Ethernet systems is one clear step toward avoiding bottlenecks and congestion. However, achieving high availability also requires an extremely fault-tolerant architecture, where no single component can fail and cause a system-wide interruption.

Also, load-sharing redundancy in key areas, such as server farms, aggregation points, and wiring closets, allows for a cost-effective redundant solution. Utilising multi-homed or aggregated links enables the network to route around a failure instead of collapsing under it.

Network redundancy implies a backup device, ie one device backs up the other. The backup device can either be active or it can be in standby.

Redundant technologies/protocols include: Link Aggregation, Resilient Links, Spanning Tree, XRN (eXpandable Resilient Networking) and VRRP.

Using all or some of these will help to build a robust and resilient network.

Active-active redundancy can be achieved by using technologies like XRN, while protocols such a VRRP provide active-standby redundancy. It is, however, important to remember that physical cable redundancy does not imply protocol redundancy, so networks need to be designed bearing both in mind.

Where companies need to build geographic redundancy into their networks, technologies like Geo XRN allow a single network core to span across a campus up to 70km away, using single mode fibre.

Neither hardware nor software can protect servers and databases against a major disaster such as fire or flood.

However, if the network itself is destroyed, critical data must remain available and easily restored to ensure business continuity.

To recover their data, companies must distribute it among multiple servers in multiple locations, such as branch offices or offsite storage sites. Mirroring the active network infrastructure keeps data up to date at all locations without introducing bottlenecks during backups.

While offsite storage and disaster recovery sites can mean the difference between survival and failure for any enterprise, they`re particularly crucial in the healthcare industry, where keeping medical records and images current and available can literally be a matter of life and death.

Manufacturing companies, too, can avoid costly delays in productivity by storing design, order and inventory records offsite.

Even the most carefully designed and protected network will inevitably experience occasional problems, especially as it increases in complexity.

A powerful, flexible network management tool helps keep small errors, like misconfigured devices or broken links, from growing into major crashes.

By watching for minor troubles, the IT department can resolve them with minimal user inconvenience and prevent recurrences.

Monitoring the network also allows a company to spot areas of rapid growth and plan network re-designs and upgrades before congestion becomes an issue - further protecting employees` round-the-clock access to the applications they need.

A network designed for high availability is a network that will reduce downtime to minutes, or even seconds, per year. With business-critical data always at hand and bandwidth-heavy applications operating at peak efficiency, companies enjoy unhampered productivity and uninterrupted supply chains.

They also serve their customers more effectively, whether they`re processing catalogue orders, scheduling patients for surgery, or registering students for classes.

In the unlikely event of a disaster, they`re able to assure their partners and clients of their continuing ability to function even as they work to recover. In short, high network availability is not only a strategy for IT success, but a key to preserving an enterprise`s reputation and long-term viability.

Fault-tolerant hardware will keep the network running, but redundant hardware does not boost network security.

Security needs to be addressed as a separate issue to network redundancy. There are many ways and levels to secure the network.

A perimeter firewall will protect a company against outside intruders - therefore preventing them from getting into the protected LAN environment.

Devices such as embedded firewalls and personal firewalls will protect inside users from rogue employees, or outside intruders that have managed to break though the perimeter defence.