From estimtes provided and detailed in Table 1 and from other intelligent guesses at the potential number of records to be entered into the data bank, the following figures have been derived for disc requirements for each species in each region. Figures shown are in millions of characters.
| Species | Region | |||
| Asia | Africa | Latin America | Total | |
| Cattle | 2.6 | 1.3 | 2.2 | 6.1 |
| Buffalo | 1.6 | .05 | .05 | 1.7 |
| Goat | .4 | .5 | .2 | 1.1 |
| Sheep | .3 | .8 | .5 | 1.6 |
| Pig | .7 | .7 | .5 | 1.9 |
| Chicken | .7 | .9 | .9 | 2.5 |
| Duck | .15 | .09 | .09 | .33 |
| Turkey | .04 | .02 | .02 | .08 |
| Rabbit and other Poultry | .2 | .1 | .1 | .4 |
| TOTAL | 6.69 | 4.46 | 4.56 | 15.71 |
Table 2. Estimated Disc Requirements of Data bank for Genetic Resources in millions of characters.
The interpretation of Table 2 is as follows. If data from all estimated relevant source documents known to date were entered into a computer system the data would occupy less than 20 million characters. To this, one must add system overheads, say 10 percent of data, still occupying less than 20 MB. The growth rate that should be applied to that figure is difficult to estimate. Earlier reports from pilot trials indicated a strong surge in the number of documents between 1960 to 1969 and then again from 1970 to 1979. The rate appeared to be slowing down in the 1980's.However, it would seem reasonable to apply a growth rate of at least 50 percent from current trends. This implies that every two years disc requirements will double. Presumably this would not go on ad infinitum, but would apply for say, the next five years. Thus, in five years' time, total disc requirements could be of the order of 100 MB. This assumes that all data available are actually entered. It may well take five years for the current backlog to be accounted for.
This amount of disc space is well within the capabilities of several micro-computers. Other requirements of the system are not major. The consultant gained the impression that results of search requests were not required with great speed. Indeed, one day, or even a week would not be considered too long to wait for information, given the speed with which research projects proceed. Working on this assumption, then, a single-user system (possibly ungradable to multi-user at a later date) with a reasonably fast printer, say 180–200 characters per second, and good print quality (but not letter quality), is all that is required. The configuration recommended would then consist only of:
1 Micro processor and keyboard
(8 bit would suffice, but preferably 16 bit)
256 KB of memory (RAM) (Less may
be used initially but this would be minimum
expansion requirements)
1–2 floppy discs of up to 1 MB
total capacity (for initial backup)
1 hard
disc, starting with 5–10 MB
1 printer (serial or parallel)
180–200 cps
1 video display unit.
A future enhancement may be a 10–12 MB magnetic tape drive for faster backup.
A single-user system means exactly that - only one person may use the system at once. More expensive systems allow that user to perform more than one task simultaneously, but it is not felt that this is necessary. The normal mode of operation would be that one person was responsible for the running of the hardware/software and that all requests would be made through that person. He/she would print the required reports and distribute them to the people requesting them.
Two options are available in setting up this configuration as mentioned previously. All data could be concentrated at one site, or each region could control its own destiny, but using cannon hardware and software to enable quick and efficient interchange of data if required. On a purely operational level it would seem desirable to opt for the regional approach. It is assumed that the actual level of information sharing between regions is quite low. Given that, the extra benefits (detailed earlier) gained from a local operation would far outweigh any disadvantages. However, on economic grounds, there is the problem of duplication of resources:
- multiple copies of software
- multiple computers
- multiple personnel to keep the system functioning satisfactorily, both in terms of computer expertise and data analyses by genetic experts.
The costs of these resources should be looked at in more detail. Assuming that the rate of data entry as a whole would be the same regardless of whether it was entered at one or several sites, then there is no extra marginal cost involved in going from a global site to several regional sites. The same could then be said of a qualified geneticist employed to screen data before they were entered. If one geneticist spent a third of his time at each of three regional sites screening data, this would be equivalent to one geneticist employed full time at a global site. Whether the same could be said of computer experts is another question. A hardware maintenance contract may have to be underwritten for each regional site, irrespective of whether any maintenance is required. Likewise the services of a full-time software expert may be required at each site even though his services could only be partially usefully utilized. The efficiency would depend on what else is to be done.
It is highly unlikely that, once the system has been designed and installed, the services of such an expert would be required for more than 10 percent of time spent in normal duties.
Assuming that the time of personnel not employed full-time in maintaining the data bank can be underwritten by other functions, then the only replication of resources is for computer hardware and software, other things being equal.
Typical prices for a system large enough to handle global data would be:
| 1. | 8 bit processor, 192 Kb memory, single 1 MB floppy disc for backup, 19 MB hard disc, terminal | US$ | 9 000 |
| 2. | 8 bit processor, 192 Kb memory, single 1 MB floppy, 42 MB disc drive, terminal | $ | 12 000 |
| 3. | 16 bit processor, 512 Kb memory, single 1 MB floppy, 15 MB hard disc, terminal | $ | 14 450 |
| 4. | As for 3, but with 33 MB hard disc | $ | 18 450 |
| 5. | As for 4 except 1 MB memory | $ | 20 800 |
To the above would have to be added a printer at say US$ 1 000.
A 12 MB magnetic tape backup system could be added for around US$ 4 750. Thus a fully serviceable system capable of running a very intelligent database package (requiring 640 KB of memory) could cost in the order of US$ 25 000. The tape backup is by no means a necessity but would be considerably faster than backing up onto floppy disc.
If a software package was chosen that would fit into 192 Kb memory and floppy disc was used for backup, then US$ 10 000 would provide a solution that would last for at least two years but provide a convenient upgrade path.
So, for a global solution, computer hardware costs could range from US$ 10 000 to US$ 25 000 depending on the options chosen.
The main cost-saving here is reduced disc requirements. A hard disc starting as low as 5 MB would suffice for all regions with upgrade path up to at least 84 MB. Such a solution may be purchased for as low as US$ 6 000 (floppy disc backup, 128 KB memory, 11 MB hard disc) to around US$ 10000 (floppy disc, 256 KB memory, 12 MB hard disc) to around US$ 13 000 (floppy disc, 786 KB memory, 15 MB hard disc).
Magnetic tape backup would again add around US$ 4 750.
