where 3600 = number of seconds per hour
F
A = fraction of computer time available
to application programs and I/O doperations
T
mI = total processor-time
spent per input message on account of application and I/O control
programs, and character transfers
to I/O device
The simplicity of Eq. (1) is in great contrast to the difficulty
in obtaining the values of F
A and T
mI.
2.
I/O Device Limit. This
is the maximum number of input messages (
Nm2) that can
be switched per hour as determined by the nature and load on the
various I/O devices such as disc file and magnetic tapes. One can evaluate
this as follows:
Nm2
= 3600/ T
m2 ................................................................(2)
where 3600 = number of seconds per hour
T
m2 = total time spent per input
message on I/O devices such as disc file and/or magnetic tape
The solution of T
m2 requires
a generous use of probability theory, statistics, queuing theory, combinatorics,
and system logic.
3.
Output Channel Limit. This is
the maximum number of input messages (
Nm3) that can be
sent per hour as determined by the the number and types (speed and service
logic) of output channels. One can evaluate this limit as
where µ = traffic
multiplication factor
lmo = average length of an outgoing message
in characters
N
o = total number of output channels
i = ith channel
fui = fraction of the time an ith output line
can be utilized for transmission of messages
tci = character duration
on the ith output channel in seconds
Based on the above limits, one can plot the capabiliity and the performance
curves. The capability curves can enable the design engineer not only
to evaulate the system bottlenects and find means to design improved programs
or hardware to extend the capability, but also to compare several complex
systems. The performance curves can yield the grade of service a customer
is getting at various input-message loads.