WPC& 2.BR Z[.#|P"^4@@l4DDd8@8dllllllllll<a݅@  I.   X(# SubheadingSubheading0\ E A.  "^``((Z(````;S>`]]]PJSJSiiiii8888ZiiiiiiSif({Z{Z{Z{Z{ZuZ`Z`Z`Z`Z.(.(.(.(x`````x`x`x`x`r]{Zx``r]x`c`{Z{i{iuiuiuixiiiiiiiii{`{i{i{i{ixii(iiii.fPiioiiiiiZ=Z(xrxixiiilililSliliiiiix`xixixix`x`rifP8SSS88/NxxxSii8V]]iiiDDiAASSiA8SSSi>>iiffSSxSrffx8SxA"xxSxXxxS姧 S88xfxxxxxxxxxx8S{ici{P8ix]i`xrxxxxxxxxxxSxxxxxxxofxGcxxxxxxxxSxxxxxxxJxxxxJxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx8xxx8xxx8xxx8xxxxxxxxxxxxx{`ic]S{``SfMx`f.+oS{Scx]xP`fc`flMiMrcxx]x{`x8irr`lrx88iiii{xiiirrr8fSJ8x Y X` hp x (#%'0*,.8135@8:100 bytes were the dominant traffic for P and 12, and the entire type of traffic for 4 and 5. DDGRespondents P, 4 and 5 showed no difference between the maximum 1 minute and 1 hour rate at 100 bytes and similar ratios for 16100 bytes. The ratio of maximum minute to typical hour was near 5 for 11 and 2.5 for 12. The ratio of typical8%).@+@+::*E@+@+::8 to maximum hour was 15 for P , 48 for 4, 1.8 for 5, 2 for 11 and 1.33 for 12. Only P reported a level of peak traffic that might approach the capacity of one LAN:  2400*80 + 300*5000 = 1.7 Mbytes = 13.5 Mbits/minute = 0.226 Mbits/sec 11 noted that a peak interval of 5 minutes was more appropriate than 1 minute.  H  2.5 Stationoriginated  H Transaction Traffic Proportion  Traffic entirely within the department was reported to be 100% by respondents 4, 6 and 9  J( and 90% by P. No respondent reported station J station traffic. The indepartment traffic was either to a host or a server. Traffic out of the department was reported at 10% by P of which 4% was offpremises. Offpremises traffic, was reported at 50% by 5, 90% by 9, and 75% by 12. All respondents reported the offpremises traffic was entirely to a host except P who noted 4% of the total traffic was to a server. 11 reported 100% of the offpremises traffic was to a host and a 50/50% division between host and server for onpremises traffic where the distinction between department and premises could not be made. It was evident that the question was interpreted inconsistently by different respondents, though this has little effect on the qualitative conclusions. Some systems are satellites of a distant host and others are primarily, but not exclusively autonomous. Both needs are present. Heavy use of an offpremises host or a remote server is consistent with methods using single destination links rather than ports on a switch or offpremises LAN taps.  H|  2.6 Comments on Other Traffic Loading  5 reported "spanning tree" protocol overhead at 510%, a significant observation. 9 reported use of diskless workstations which download 12 Mb applications from a server when booted.  H#  2.7 Load from Broadcast Traffic The loading from broadcast traffic was included in the report in 2.3 for P, 4, 6 and 9, but not for the others. Broadcast loading was reported as over 5% of the traffic by P, 5 and 6. (.@+@+::  H  2.8 Traffic Timing Constraints  DDGOnly 5 respondents made any attempt to complete this section, and none addressed more than a fraction of the blanks. DDGRegistration at entry was commonly allowed 35 seconds and 30 seconds by 11. DDGAccess delay was not broken out between typical and worst case by any respondent, and the separation of known present and projected requirements drew no separate distinction. It is concluded that these respondents do not see requirements different from the present performance. DDGAccess delay values, where given, showed values of 0.5 and 5 milliseconds (maybe) and 1 and 5 seconds. It is believed that the large values are associated with offpremises links and include delays unrelated to the LAN itself. The small values, if they are not a confusion in units, are those associated with predominantly indepartment traffic. DDGTransit delay values  appeared equally obscure. 11 alone seemed to have a sense of the difference between access and transit delay requiring 5 milliseconds access delay and allowing 45 seconds for transit delay. 11 alone gave variability of transit delay at 23 seconds. 4 had no different answer for access and transit delay. 12 had a longer provision for transit delay, but both were in seconds. DDGAn acknowledgment requirement was seen only by 12 and that presumably was an upper layer function that already exists. P gave a 2 second wait for ACK as a requirement. In all other cases, no information was offered.  H   2.9 Connectiontype Service  DDGOnly 5, 9 and 11 responded with a requirement which included the analog modem function for all and coded voice for 5 and 11. DDGLittle direction was obtained from this question.  H!  3.0 OPERATING ENVIRONMENT   H(#  3.1 Operating Environment Description  DDGThe environments that were most frequently checked were rooms, hallways with fixed partitions, high ceiling cluttered interior, single floor building. Only two checked suspended acoustic ceiling and three open areas with movable partitions. 4 only checked outdoor area descriptions.8(.@+@+::l)SS@+@+::8 Boundary  conditions were checked by P, 4 and 11. 11 only noted space isolation. P and 11 noted liketype areas with P giving proportions.  P in 40% of the cases and 11 noted that the service boundary and walls coincided,  H  3.2 Quantities and Dimensions  Hx for Wired and Wireless  No. stations per enclosed area were small numbers in the range of 1 to 15, except for 5 who reported 1000, for the maximums. The typical numbers were 57 except for 5 where it was 200. The enclosed area was typically given as 12, 30 or 1600 meters by the three respondents. Both P and 5 are large systems but P is divided into much smaller rooms.  HH  3.3 Quantities and Dimensions  H for Wireless Only with Infrastructure  Only 4, 11 and 12 responded in this sections. 11 gave only number of users and LANs and no information on the size of space.  ~J0 4 and 12 both showed near 10 meters per  ~J user typical for 20 to 100 meters of enclosed area. The per premise answers were essentially the same as answers for one enclosed area for 4 and 12.  H  3.4 Quantities and Dimensions  Hp for Wireless BCA with No Infrastructure  P, 4 and 6 attempted answers. 4 repeated answers for one enclosed area in 3.2 for one BCA in this question. P gave values that were equivalent to a few enclosed areas in 3.2. 6 reported values in which 27 autonomous wireless stations were operating within a much larger environment. Station density was not much different from previous answers by P and 4. P alone understood that several basic coverage areas (BCA) might compose one total coverage area (TCA) giving 10 BCAs per TCA as a maximum.  H(#  3.5 Colocated LANs  Colocated LANs are considered likely by 5 and 11, possible by P, 4 and 9 and improbable by 6 and 12. A common minimum distance between centers of colocated LANs was give as 40 meters by P and 4 and nil by 5 and 9. 11 said colocated LANs would be merged. (.@+@+::  J  4.0 MOBILITY REQUIREMENT  DDGMajor incentive to use wireless was reported for staff away from desk only by 11, for wiring inaccessible places by P, 6 and 11, for quick relocation moves by P, 11 and 12. Nearly all categories of incentive suggested were checked as a minor incentive by 11. DDGThis group or respondents is very light on mobile applications and strong on wiring alternatives for cost or convenience.  H  ANALYSIS AND INTERPRETATION  DDGIt is clear that the drafting of the questionnaire was not well matched to the responding group's knowledge or operation, and it did not elicit more than a fraction of the information that was sought. After seeing the answers, many of the questions could have been differently drafted. DDGOne or two did not read the introductory text for definitions, and so many blanks were unfilled that the form must be faulted. DDGIt is also seems that some of these users do not have any way of knowing the breakdowns on traffic types in their own networks, and most did not choose to guess. Perhaps we are hoping for too much in obtaining designer's information by this means. DDGThe information obtained on delay was superficial and not well related to source cause. DDGThe sample is representative of the categories responding which are mostly users of mainframe services. The autonomous networks used for engineering and manufacturing do not appear. The absence of stationstation traffic is striking. DDGSome good information was obtained that supports user density assumptions near 1 per 10  ~J" meter and the existence of room sizes with enough stations to support one or more accesspoints. DDGThere is not enough data to support exclusion of cases not reported or reported unused. DDGRespondent 4 intended a wireless LAN, and gave decent information about the environment, and on typical and peak transaction rate for over 100 byte traffic entirely within a department. DDGExcept for 4, these respondents looked at wireless as a wiring replacement rather than a means of enabling station mobility. It is an indication that a large part of the wired LAN using community may only have a casual interest in wireless for a small number of situations where the wiring is inconvenient.