ATT DEFINITY Generic 3 Call Vectoring/Expert Agent Instructions Manual

							Looping Examp les
Issue  4 September 1995
M-5
The example in Figure M-5 a d ds a delay so that the lookahead interflow attempt 
occurs only every 10 seconds.
Figure M-5. Example Vector with Improved Performance
The example in Figure M-6 increases performance even more by increasing the 
d elay between lookahead interflow attempts to 30 se conds.
Figure M-6. Another Example Vector with Improved Performance
Table M-2 c ompares the relative processing cost of the three examples by 
looking at the a pproximate number of vector steps executed while processing 
the call. Assump tion is that the announcement is 5 seconds long.
When a call is queued for 5 minutes, the amount of vector steps drops 
d ramatically when a delay is a dd e d (Figure M-5), and drops even more when the 
length of the delay is increased (Figure M-6). When  an  agent in split 1 is 
imme diately available to answer the call, there is no difference in the amount of 
vector steps for the three examp les. Table M-2. Approximate Number of Vector Steps Executed 
for Lookahead Interflow Examples
Example in 
Figure M-4Example in 
Figure M-5Example in 
Figure M-6
when a agent is available in 
s plit 1111
q ueueing time of 5 minutes up to 1,000 85 30
1. queue-to main split 1 pri l
2. announcement 3000
3. wait-time 20 seconds hearing music
4. route-to number 93035555555 cov n if unconditionally
5. wait-time 10 seconds hearing music
6. goto step 4 if unconditionally
1. queue-to main split 1 pri l
2. announcement 3000
3. wait-time 20 seconds hearing music
4. route-to number 93035555555 cov n if unconditionally
5. wait-time 30 seconds hearing music
6. goto step 4 if unconditionally 
						

							Improving Performance
M-6Issue  4 September 1995 
Check Backup 
Recommendation: When using check-backup commands to queue a call to 
backup splits, ensure that an adequate amount of time has elasped before 
checking the b ackup splits again. 
NOTE:
With the introduction of the ‘Expected Time Wait Time’ feature in Generic 3 
Version 4, the style of programming used in this example is no longer 
relevant. The  best a p proach is to use the ‘Expected Time Wait’ feature to 
locate the most appropriate split for the call and queue it there.
The example in Figure M-7 checks backup splits continuously as long as the call 
is in queue.
Figure M-7. Example Vector 
The examp le in Figure M-8 adds a delay of 10 seconds to ensure that some time 
has elasped before checking the backup splits again.
Figure M-8. Example Vector with Improved Performance
1. queue-to main split 1 pri h
2. announcement 3000
3. wait-time 10 seconds hearing music
4. check-backup split 21 pri m if available-agents > 0
5. check-backup split 22 pri m if available-agents > 0
6. check-backup split 23 pri m if available-agents > 0
7. check-backup split 24 pri m if available-agents > 0
8. check-backup split 25 pri m if available-agents > 0
9. goto step 4 if unconditionally
1. queue-to main split 1 pri h
2. announcement 3000
3. wait-time 30 seconds hearing music
4. check-backup split 21 pri m if available-agents > 0
5. check-backup split 22 pri m if available-agents > 0
6. check-backup split 23 pri m if available-agents > 0
7. check-backup split 24 pri m if available-agents > 0
8. check-backup split 25 pri m if available-agents > 0
9. wait-time 10 seconds hearing music
10. goto step 4 if unconditionally 
						

							Looping Examp les
Issue  4 September 1995
M-7
Since the agent availability status may not be likely to change every 10 seconds, 
it may make sense to increase the wait time to 30 seconds, as shown in the 
example in Figure M-9.
Figure M-9. Another Example Vector with Improved Performance
Table M-2 compares the relative processing cost of the three examples by 
looking at the approximate number of vector steps executed while processing 
the call. Assumption is that the announcement is 5 seconds long.
When a call is queued for 5 minutes, the amount of vector steps drops 
dramatically when a d elay is a dd e d  before checking the backup splits again 
(Figure M-8), and  drops even more when the length of the delay is increased 
again (Figure M-9). When an agent in sp lit 1 is immediately available to answer 
the call, there is no difference in the amount of vector steps for the three 
examples. Table M-3. Approximate Number of Vector Steps Executed 
for Check Backup Examples
Example in 
Figure M-7Example in 
Figure M-8Example in 
Figure M-9
when a agent is available in 
split 1111
queueing time of 5 minutes up to 1,000 190 65
1. queue-to main split 1 pri h
2. announcement 3000
3. wait-time 30 seconds hearing music
4. check-backup split 21 pri m if available-agents > 0
5. check-backup split 22 pri m if available-agents > 0
6. check-backup split 23 pri m if available-agents > 0
7. check-backup split 24 pri m if available-agents > 0
8. check-backup split 25 pri m if available-agents > 0
9. wait-time 30 seconds hearing music
10. goto step 4 if unconditionally 
						

							Improving Performance
M-8Issue  4 September 1995 
Other Examples
After Business Hours
Recommendation: Test to see if the d estination resourc es are available (such as 
during business hours) before queuing.
The example in Figure M-10 q ueues calls to a hunt group regardless of the time 
of the call. When the call is made after b usiness hours, the announcement is 
repeated until the caller hangs up.
Figure M-10. Example Vector 
The example in Figure M-11 tests for business hours before queuing the call. If 
the call is made after business hours, an announcement informs the caller of the 
business hours and the call is terminated.
Figure M-11. Example Vector with Improved Performance
In the first example, unnecessary processing occurs when a call is queued after 
business hours and the call is terminated only when the caller hangs up. As 
shown in the second example, it is more economical to test for b usiness hours 
before queuing a call.
1. queue-to main split 1 
2. announcement 5000 
(“All agents are busy. Please hold.”)
3. wait-time 120 seconds hearing music 
4. announcement 5001 
(“All agents are still busy. Please continue to
hold.”)
5. goto step 3 if unconditionally 
1. goto step 7 if time-of-day is all 17:00 to all 8:00
2. queue-to main split 1
3. announcement 5000 
(“All agents are busy. Please hold.”)
4. wait-time 120 seconds hearing music
5. announcement 5001 
(“All agents are still busy. Please
continue to hold.”)
6. goto step 4 if unconditionally
7. disconnect after announcement 5001 
(“Business hours are 8:00 AM to 5:00 PM,
Please call back then.”) 
						

							Other Examples
Issue  4 September 1995
M-9
Lookahead Interflows
Recommendation: When  using  a  lookahead interflow, first test to see if the 
receiving office is open for business.
The scenario is a sending switch in Los Angeles, with office hours from 8:00 AM 
to 5:00PM (8:00-17:00) PST and the receiving switch is in New York, with office 
hours from 8:00 AM to 5:00PM EST (5:00-14:00 PST).  There is a 3 hour d ifference 
between the two switches. 
The example in Figure M-12 routes calls to the New York switch. If there are no 
agents available at the Los An geles switch. It is possible for calls to be 
interflowed d uring hours that the agents in New York are not available, thus doing 
unnecessary processing. 
Figure M-12. Example Vector 
The example in Figure M-13 tests first to see if the New York switch is open 
before requesting a queue to the New York switch, thus avoiding unnecessary 
processing.
Figure M-13. Example Vector with Improved Performance
1. queue-to main split 1 
2. route-to number 99145555555 cov n if unconditionally
3.  announcement  2770  (“All agents are busy. Please hold.”)
4. wait-time 120 seconds hearing music 
5. goto step 3 if unconditionally 
6. stop
1. queue-to main split 1 
2. goto step 4 if time-of-day is all 14:00 to all 05:00
3. route-to number 99145555555 cov n if unconditionally 
4.  announcement  2770  (“All agents are busy. Please hold.”)
5. wait-time 120 seconds hearing music 
6. goto step 4 if unconditionally 
7. stop 
						

							Improving Performance
M-10Issue  4 Septemb er 1995 
The example in Figure M-14 can b e used if you have Generic 3 Version 4 
Advanced Routing  optioned. In this case, the ‘Expected Wait Time’ feature may 
be used to determine whether it is worthwhile placing a lookahead interflow call 
attemp t.
Figure M-14. Another Example Vector with Improved Performance
There is little reason to attempt an interflow if the call will be answered quickly at 
the main switch. For the exam ples in Figure M-13 and Figure M-14, vector steps 
are avoided which do not aid in the call being answered sooner.
1. queue-to main split 1 
2. goto step 5 if expected-wait for call < 30
3. goto step 5 if time-of-day is all 14:00 to all 05:00
4. route-to number 99145555555 cov n if unconditionally 
5.  announcement  2770  (“All agents are busy. Please hold.”)
6. wait-time 120 seconds hearing music 
7. goto step 5 if unconditionally 
8. stop 
						

							Relative Processing Cost of Vector Command s
Issue  4 September 1995
M-11
Relative Processing Cost of Vector 
Commands
Some vector commands use more p rocessing resources than others. Table M-4 
and Table M-5 show the relative processing costs of specific vector commands 
for Generic 3 Version 4i/v/vs and Generic 3 Version 4r respectively. Whenever 
possible, use the lower cost vector commands. This will minimize your 
performance costs and upgrade your performance.
Table M-4. Relative Processing Cost of Vector Commands for 
Generic 3 Version 4i/v/vs
relative 
performance
 cost vector command
high adjunct routing
high c heck-backup 
high c ollect  digits
high q ueue-to
high route-to 
high c ollect < 1 > digits
medium c onverse
medium announcement 
medium g oto step 
medium g oto vector 
medium messaging
low b usy
low d isconnect 
low sto p
low wait-time 
						

							Improving Performance
M-12Issue  4 Septemb er 1995 
Table M-5. Relative Processing Cost of Vector Commands for 
Generic 3 Version 4r
relative 
performance
 cost vector command
medium adjunct routing
medium c heck-backup
medium c ollect digits 
medium c onverse 
medium g oto vector (table comparison) 
medium messaging 
medium queue-to 
medium route-to 
low announcement 
low b usy
low d isconnect
low g oto step 
low g oto vector 
low sto p
low wait-time 
						

							Issue  4   Septemb er 1995N-1 
N
Call Vectoring System Parameters
G3iV1.1 and G3V2 System Parameters
1. The total of VDN, Station, and Login ID extensions cannot exceed 25,000.
2,000 for G3rV3
7,084 for G3rV3 for Simultaneous 3-way Conf. Calls
Table N-1. Maximum System Parameters for Call Vectoring/EAS 
for G3V2
ITEMG3iV1.1-
286G3vsV2 
ABP/PBPG3sV2 
ABP/PBPG3iV2-
386 G3rV2
Call Vectoring
Multiple Splits per Call 3NA/3 NA/3 3 3
Priority Levels 4 NA/4 NA/4 4 4
Recorded Announcement 128 NA/128 NA/128 128 256
Ste ps pe r V e c tor 32 NA/ 32 NA /32 32 32
Vector Directory Numbers
1500 NA/100 NA/100 512 20,000
Measured VDNs 500 NA/100 NA/100 512 2,000
Vectors per System 256 NA/48 NA/48 256 512
Expert Agent Selection (EAS)
Skill Groups NA NA/24 NA/24 99 255
VDN Skill Preferences NA NA/3 NA/3 3 3
Multiple Skills p er Call NANA/3 NA/3 3 3
Multiple Skills p er Agent 4NA/4 NA/4 4 4
Agent Login IDs NA NA/450 NA/450 1,500 10,000 
						

							Call Vectoring System Parameters
N-2Issue  4   September 1995 
G3V3 System Parameters
1.The total of VDN, Station, an d Log in ID extensions cannot exceed 25,000.
2,000 for G3rV3
7,084 for G3rV3 for Simultaneous 3-way Conf. Calls
2. Measured limits depend on the CMS release used.
Table N-2. Maximum System Parameters for Call Vectoring/EAS 
for G3V3
ITEMG3vsV3 
ABP/PBPG3sV3 
ABP/PBP G3iV3 G3rV3
Call Vectoring
Max. Skills a Call Can Simultaneously
Queue toNA/3 NA/3 3 3
Priority Levels NA/4 NA/4 4 4
Recorded Announcement NA/128 NA/128 128 256
Steps per Vector NA/32 NA/32 32 32
Vector Directory Numbers
1NA/100 NA/100 512 20,000
CMS Measured VDNs
2NA/100 NA/100 512 2,000
Vectors per System NA/48 NA/48 256 512
No. of Collected  Digits for 
Call Promp tingNA /16 NA/ 16 16 16
No. of Dial-Ahead  Digits for 
Call Promp tingNA /24 NA/ 24 24 24
Vector Routing Tables NA NA NA NA
Expert Agent Selection (EAS)
Skill Groups NA/24 NA/24 99 255
VDN S kill  Pre fe ren c e sNA/3 NA/3 3 3
Max. Skills a Call Can Simultaneously
Queue toNA/3 NA/3 3 3
Agent Login IDs NA/450 NA/450 1,500 10,000
Max. Skills per AgentNA/4 NA/4 4 4
Max. Agents that can be Logged-In
When Ea ch  Has  4  Skills  Assig ne d NA/37 NA/37 125 1,300
When Ea ch  Has  1  Skill AssignedNA/150 NA/150 500 5,200