Not all of the technology that we use today is advancing at the same rate. It is I think likely that it is the diferences in development rather than any fundamental new developments that will shape computer use over the near future.
In recent years performance has been gained by transfering data between a higher performance static processor memory cache and the main Dynamic memory in larger chunks. This has been increased from 1 byte at a time, first to 2 bytes and more recently to 4 bytes at a time.
Some of the latest server systems have 64bit {8 byte} or even 128bit {16 byte} data busses for transfering information. But when most information is textual and based on single byte charecters this getting in to the realms of diminishing returns.
If you make 10,000,000 memory transfers per second each of 8 bytes That give you 80MB transfer per second.
Their are curently only two real options for further performance gains. Switch to using only static memory for the main system memmory. This has a cost penalty but can be as much as 6 times faster. Also it uses less power. Also a lot of curently available software has grown with the development of the PC and doesnot realy seem to make the best use of the memmory available. Is a modern text editor that ocupies 16MB realy a 1000 times better than say the original CPM wordstar program that ran in 16KB.
Falling prices are likely to bring 100MHz swiched ethernet, 125MHz ATM, and Gigabit Ethernet down to this sort of prices in the next few years.
However a point that may have a major bearing on the development of the LAN is that data being transfered has to be coming from somewhere, 10MHz swiched ethernet will pransfer around 800KB per second each way, which is well over the random access performance of a hard disk.
100MHz swiched ethernet at 8MB per second each way, or 125MHz ATM at around 12MB per second, both exceed the linear access performance of a hard disk or even of a striped disk array.
If switched Gigabit or the next incriment of ATM were implimended to the desktop they would both exceed the data band with of current and projected memory systems. See memory above.
Technology moves on with the new buzz word beeing DVD or Digital Versatile Disk. This change in format increased the amount of data that can be stored on a 120mm disk to 4.5GB, with the promise of 18GB for a double sided multi layer disk in a few years time. Thats more than enough for the longest film even if stored for playing at high definition say 1600x1280 pixels. With sound tracks for every country on earth.
Now given that the typical Novel {no pictures} is about 2MB, in computer storage terms a picture is equal to more like 50,000 words that the 1000 of the old saying. The time is rapidly approching when you could put a world Encyclopidia, A complete teliphone directory of everyone on earth, A compleate Geniology of everyone who has lived in recorded history, and more books that anyone could read in a liftime, on a handfull of silvery disks that will fit in your pocket.
Since the Cathode Ray Tube {CRT} was invented the technology has improved. The number of pixels {dot triplets} that can be used to make up the display has increased substantialy. This can be seen with computer displays which have improved from the original 14" CGA displays at 320x200 pixels, 25Hz interlaced. To the curent "Standard" display 17" 1280x1024 pixels, 75Hz non interlace.
Substatial improvements in the quality of the display is available, whether the signal source is a digital transmition or an analog one. By storing the received image in memory the same as on a computer display. And then driving one of the computer grade CRT's at 75Hz. If you store 2 or three frames from an analogue source with only a a relatively limited amount of signal processing you can eliminate most of the background noise.
The asymetric bit in the name comes from the fact that the speed is slower in one direction than the other.
If you are close to an exchange and can successfuly get a modem to operate at 56k/32k on you ordinary teliphone line. Theoreticly ADSL would give you 1.5M/512k receive and transmit respectively.
This is very similar to speeds attainable with the "Cable Modem" services offered by some of the cable TV companies.
For availability in the UK check with BTopenworld
General note BT is aiming to provide a feed to each exchange covered based on 1 20th of the nominal band width of the ADSL links provided. So for an exchange with 40 512kbps ADSL links installed this would equate to a backbone link of 20 x 512 / 20 or 2Mbps.
The PAD's maintain the virtual links between the serial tail circuits. They also make shore that the data is reconstructed in the correct order, and that any packets that are droped or corrupted are retransmited.
These days Frame relay, ATM, or dialed ISDN, solutions are almost cirtainly more appropriate.
User links are normaly at speeds of between 64kbps and 8Mbps, packets are sent across virtual circuits between predetermined endpoints.
Error detection and retransmition is however down to the users equipment. As it is a shared transmition network, end to end transmition time {Latency} depends on how busy the network is with other peoples traffic rather than just your own.
If you are in a few countries such as Germany the telicoms operators allow you to access the framerelay network via the spare band width on the control chanel of an ISDN circuit.
This allows routers with the capability to use this service to maintain continueous low bandwidth connectivity to all your sites, bringing up a chargible ISDN call only when data volumes justify it.
From a position only a yearago when the Telco's would realy only recomend/supply frame relay if you wanted to connect 10+ sites in a complete mesh, or to link multiple countries. With charges based on the volume of data transfered.
You can now get framerelay links with as few as 4 sites on a fix anual fee basis. This partly reflects two changes. As techonogy has improved the band width available on the installed fiber has been improving at the rate of around 40% per year, As the majority of the cost in setting up a fiber network is installing the fiber in the first place this means that the cost of bandwith to the telco's should have been falling at a simmilar rate.
Also as major companies switch from framerelay to ATM for their future requierments, the telco's have spare capacity on their framerelay networks.
As well as accepting asyncronous variable volume data streams on an if bandwidth available system. A user device can request a fixed bandwith temport or semi permanent circuit just like an ISDN router bringing up 1 or more ISDN B, H , or J chanels according to the amount of bandwith requiered.
The other difference is the standards were designed from the start to scale into the multi gigabit range and beyond.
This combination of feature make it ideal for telecoms operators and large companies who wish to mix Permenent circuits for applications such CCTV monitoring, semipermanent fix bandwidth Phone or vidio-conference calls and continuously varying network traffic on the same infrastructure.
This avoids the need to dig up the road to lay cables. Many people may have come across this as used by Ionica.
Motorola and cisco systems have resently set up a joint venture "SpectraPoint Wireless" to purchase the Bosch Telicoms infrastructure, with the intention of further developing the technology.
See artical in Electronics times 14 June 1999 for more details.
However a single base station could cover a radius of upto 20 miles. With may be 20 simultainious calls within the area covered.
This made it relatively cheap to provide a national network. as long as you did not have too many subscribers.
The second generation of the mobile network was digital. With an availble digital chanel of upto 9600 bits/s providing you had a reasonable strong signal. Also the radius covered from a base station was much smaller typicaly around 5 miles. This means that far more base stations are needed but with increasing popularity this is an advantage as at a similar number of calls per base station; "more stations" = "more calls" = "more customers"
The digital nature of the 2nd generation mobile network has allowed other services to be developed. With handsets such as the Nokia 9000 series offering SMS, FAX, Telnet, and limited web browsing.
The race is now on to develop the standards and chips to support a third generation mobile network offering higher bandwith, better quality voice and, the posibility of additional digital network services.
In persute of this target Sirius Comunications, a subsidury of the Belgian inter-university microelectronics reasearch institute {IMEC}. Has made available for licencing a chip design for a programable decoder to work with the proposed standards.
A test chip is being fabricated by Alcatel incorporating this decoder and an ARM7 microprocessor on a single low power 240pin chip. See artical in Electronics times 14 June 1999 for more details.
Given the moves mentioned above to combine an ARM7 processor with the G3 mobile phone network circitry. It would be perfectly practical to combine the functionality of an NC and a mobile telephone.
My home is linked via fiber or an ADSL line to a local hub from where I receive vidio steams on demand, Internet access and link with an ATM uplink to one of perhaps half a dozen super exchanges covering the UK. This ATM link allows me to dial 64K voice, or 384K Vidio conference calls to anywhere in the world. At a lower cost than for current voice calls.
The display is a large flat screen hung on the wall which when not in use for TV, vidio phone, or network access can display an image of any picture I like.
When I am away from home, and a visitor calls. My house sends a message to my mobile workstation vi its third generation mobile comunications link. I can then call up a live view of the visitor, and speek to them if I so wish. Further if I so wish I can instruct the house to let the visitor in to all or part of the premisis. of the premisis.