On Wednesday 7 August this year, America Online crashed and stayed down for 19 hours, locking all six million of its members off the Internet.
On 19 June, Netcom, the fourth largest US ISP, crashed for 13 hours. Bob Metcalfe, inventor of Ethernet, the local area networking (LAN) standard, recently wrote in his column in Infoworld: "I predict the Internet will soon go spectacularly supernova and in 1996 catastrophically collapse."
The Internet has undergone a remarkable transformation. The original Internet was a low-speed, text-based network used to connect a few government sites to the research and defence contracting community. In a few months' time, the Internet backbone will be operating at speeds which were inconceivable a few years ago. Yet there is concern that even these will not be fast enough to alleviate the Internet's worsening traffic congestion.
Limited bandwidth, increasing traffic and greater capacity requirements to support large files such as multimedia are creating bottlenecks which make it agonisingly slow or impossible to tap into some sites at peak times.
The effect of Internet congestion on users has in turn made an impact on the technology. Trends for the future point to the Internet backbone network being converted from its current T3 design (operating at 45Mbps) to Asynchronous Transfer Mode (ATM) at speeds of 155Mbps today and 622Mbps in the near future.
Understanding the issue of capacity requires a basic knowledge of how the Internet works and the complex process which enables information transfer.
The thousands of networks which make up the Internet communicate using two protocols - Transmission Control Protocol (TCP) and Internet Protocol (IP). In simple terms, TCP breaks down information into packets (envelopes of data) for transfer and reassembles them at the destination point. IP ensures proper delivery of data to the right address.
Many stops are made between sending and receiving, including hubs, bridges, gateways, repeaters and routers. Hubs link groups of computers and bridges link LANs. The function of gateways, similar to that of bridges, also involves translating information from one network to another.
Because data often travels a long distance, repeaters ensure continued signal strength. Routers manage the traffic, sending packets from router to router towards their destination. A chain of routers, each of which knows the address of the other routers on the Net (thanks to constantly updated router tables) passes the packets along to their final stop.
The huge growth of Internet traffic has required a concomitant increase in routeing power. Last autumn, Cisco (which has 90 per cent of the routeing market) introduced the Model 7500 router, with a data processing power of 2Gb per second - that's a fourfold increase over its high-end predecessor, the Model 7000. Cisco plans to double capacity again this year but even that may not be enough. And controversy still rages over whether IP routers will be able to handle the Internet's burgeoning traffic or whether ATM switches should replace them.
Unlike IP routeing, ATM is a point-to-point system - that is, a "virtual circuit" is established from end to end and kept in service until a session is closed. ATM is especially well suited to realtime multimedia applications in which it is critical that data arrives in a timely fashion. Unlike IP packet routeing, ATM cells all follow a predictable path once the circuit is established.
IP router proponents argue that routers can handle the traffic are more flexible in their ability to route packets along different paths, whereas ATM is a rigid system requiring fixed connections. ATM supporters counter that ATM switches are faster and ATM networks can be made flexible through segmentation.
Today, ATM is employed mainly by large network providers for their backbones.
Only time will tell whether ATM switching will be used exclusively for long-haul links or whether the technology will migrate down to local levels.
Could be used for everything from search-and-rescue robots to wearable tech
Don't require the rare material being mined from the mountains of South America
IBM hopes that its new tool will avoid bias in artificial intelligence
Found by calculating the strength of the material deep inside the crust of neutron stars