Capacity Planning
|
 |
 |
|
 |
 |
Benchmarking Conclusions
This section provides illustrations of benchmarking conclusions based on tests performed:
- Solaris Platform Summary - Linear Scalability
- Solaris Platform Summary - Horizontal Scalability
- Windows Platform Summary - Horizontal Scalability
Solaris Platform Summary - Linear Scalability
The following conclusions were reached based on linear scalability (increasing the number of processors on the same box).
Figure A-1 Light-weight (MedRec) Using HTTPS
- The throughput of a two-processor configuration is 1.79 times that of a one-processor configuration.
- The throughput of a four-processor configuration is 1.80 times that of a two-processor configuration.
- The throughput of an eight-processor configuration is 1.52 times that of a four-processor configuration.
Solaris Platform Summary - Horizontal Scalability
The following conclusions were reached based on horizontal scalability (increasing the number of boxes, using a clustered configuration).
Figure A-2 Light-weight (MedRec) Using Clustered Configuration
The throughput of a one-node to four-node cluster shows linear scalability.
Windows Platform Summary - Horizontal Scalability
The following conclusions were reached based on horizontal scalability (increasing the number of boxes, using a clustered configuration).
Figure A-3 Light-weight (MedRec) Using Clustered Configuration
- The throughput of a two-node configuration is 1.85 times that of a one-node cluster.
- The throughput of a three-node configuration is 2.82 times that of a one-node cluster.
- The throughput of a four-node configuration is 3.72 times that of a one-node cluster.
- HTTPS test on 4X400 Mhz Solaris configuration gives 60% performance of HTTPS test.
- HTTPS test on 4X750 Mhz Solaris configuration gives 58% performance of HTTPS test.
- HTTPS test on 4X700 Mhz Windows 2000 configuration gives 49% performance of HTTPS test
Figure A-4 Heavy-weight (MedRec) Using HTTPS
