Expand Configuration and Management Manual (H06.21+, J06.10+)

Tuning
Expand Configuration and Management Manual — 529522-013
19 - 2
Performance Factors
Performance Factors
This subsection describes the factors that can be adjusted to improve Expand line-
handler process performance and processor utilization. Performance factors, and their
relative effect on the tuning goals described earlier in this section, are shown in
Table 19-1.
How to Use the Performance Factors Table
To use Table 19-1, vertically scan the Tuning Goals columns. A value of 1 indicates the
greatest effect or impact and is thus the first factor you should examine when
attempting to achieve a specific tuning goal; a value of 6 indicates the least impact. For
example, if you are attempting to optimize resource use, you will look at multipacket
frame size, variable-packet size, and application message size first and Layer 2
window size last.
If more than one performance factor is given the same priority (for example,
multipacket frame size, variable-packet size, and application message size are all
valued at 1), this is an indication that the factors are interrelated and should be
examined simultaneously.
The performance goals listed in Table 19-1 are described in detail in the remainder of
this subsection.
Table 19-1. Performance Factors
Performance Factors
Tuning Goals
Resource Use/Cost Throughput Network Delay
Multipacket Frame Size 1 2 3
Variable Packet Size 1 1 3
Application Message Size 1 1 1
Packet Format 5 4 3
Congestion Control 5 1 1
Layer 2 Window Size 6 4 5
Processor Type (See Note 1) (See Note 1) (See Note 1)
NAM Interface 2 2 1
Data Compression 4 3 4
Multi-Line Paths 3 3 3
Multi-CPU Paths 3 3 4
Network Topology (See Note 2) (See Note 2) (See Note 2)
Note 1. Faster processors provide faster response time and higher throughput, given sufficient bandwidth.
Though this factor has a very large impact on the noted goal, it is the most costly or difficult factor to change.
Note 2. Network topology, particularly the location of passthrough nodes, can affect resource use, throughput,
and response time.