The matrix: Mesh networks catch on
In view of the increasing volume of data, the ring structures common today are reaching their limits. At the same time, modern networks must be able to react flexibly to new designs. But considering the innumerable protocols, incompatible components, and high investment costs, the providers of data transmission services are finding it difficult to balance these needs. Mesh optical networks represent the solution to this dilemma.
Ring structures have become the dominating network topologies, particularly in the networks of towns. They have one great advantage: If one section of the ring should be interrupted, traffic can be rerouted via the intact segment. But the other side of the coin is that all the hardware components used to transmit and forward data must be designed as dual systems. That is all the more expensive the more network nodes are integrated in a ring. Should an additional malfunction occur, the self-healing properties of the ring are virtually exhausted. For a long time, however, these disadvantages made no difference.
Overall, the share of voice traffic in the networks has fallen sharply in the last few years, while data traffic has increased dramatically and is still growing particularly because of the Internet. Data traffic caused by the Internet is currently doubling every six to 12 months—a trend that all levels of the network are up against, from LANs to long-distance networks. However, more and more globalised companies are also entering the metro sector, driven by their own economic developments.
In some cases, the maximum data rate in corporate SANs is increasing every year. Speeds of 1 Gbit/sec are quite common, and we expect they will soon rise to 2 Gbits/sec. Within two to three years at the latest 10 Gbits/sec will be generally available.
While most attention used to be paid to purchase costs, the focus has clearly shifted to operating costs in the IT/SAN sector. Nonpayment risks can be reduced drastically and at the same time running costs can cut through outsourcing.
The variety of protocols in the computer-centre sector is growing, leading to an increase in the variety of hardware and software. That in turn leads to an increase in the effort and expenditure involved in the design, restructuring, and consolidation of SANs.
Since many companies are dependent on the proper functioning of their company software and at the same time have to archive their data for, say, 10 years, a reliable storage-area network is critical for the existence of the company. The consequence is that a large amount of money and effort is required to archive data, and at the same time the data networks must be backed up redundantly.
Globalisation is not only creating growing mountains of data, but it's also speeding up the lifecycle of network structures, forcing companies to react quickly to the constantly changing world financial situation. Modern networks have to be able to mirror this continuing structural change at shorter and shorter intervals.
Customers have to get used to the idea that they will not be able to obtain the services they require from "their" provider forever. Furthermore, they have been made nervous by the enormous variety of products and suppliers that dominate the network sector and that are conducting an increasingly fierce price war. A common reaction to this unpredictability is for customers to employ system integrators they trust and ask them to select compatible components from the range of many different suppliers. These are not the only challenges providers are facing today:
- They offer transmission services with quality of service on ATM platforms and SDH, but the effective data transmission rate achieved is less than 50% of the nominal data rate of the service sold—to the annoyance of customers.
- The usual service life of their equipment has dropped from five to 15 or even more years in the past to just a few years now.
- Redesigning the network affects not only the network topology, but also the type of applications and associated hardware components, leading to relatively high repeated investment costs.
- Despite the high cost pressure, existing customers must be retained by means of new applications such as UMTS and at the same time new customers have to be acquired.
- What's called for is a high level of future orientation and flexibly adaptable topologies involving a minimum of investment. The main focus here is on protocol transparency. Further requirements must also be met—increasingly complex networks must remain easy to survey.
Mesh networks accurately reflect the geographical conditions of the locations and the existing cable structures. The structures of future optical networks can no longer be forced into today's patterns of point-to-point connections or ring topologies. Mesh networks are already widespread and will be used more and more in the years to come. No wonder, because the use of equipment for mesh networks has advantages not only in the networks of towns, but also in the storage sector. A good example of that even for simple data mirroring to a third node is where the hardware components are used during the day for normal production and at night for backups.
In practice, mesh-network structures are often artificially forced into ring and point-to-point structures since numerous systems on the market are not in position to implement mesh networks with the existing hardware or the available network management. By contrast, units designed specifically for mesh networks provide greater flexibility in existing network structures and reduce investment costs for system hardware.
Compared to the mirroring of networks on a ring-type structure, which is very common on the market, equipment for mesh networks can allow savings of up to 70% in assemblies and up to 40% in operating costs. So from a cost point of view, there is no longer any reason to stick with out-of-date ring structures.
Elke Jahn, Chief Executive