Compaq ProLiant 1000 Performance Analysis and Tuning of Raptor's Eagle NT 3.06 - Page 15

Base System, Memory

Get Compaq ProLiant 1000 PDF manuals and user guides View all Compaq ProLiant 1000 manuals
Add to My Manuals
Save this manual to your list of manuals

Page 15 highlights

278A/0497 WHITE PAPER (cont.) ... subsection contains some test runs based on lower end ProLiant and ProSignia systems and was run to show performance on lower-end systems. Base System The base system, test run 1, consists of the ProLiant 5000 system, 1Pentium Pro 200 MHz, 512K cache processor, 64 MB RAM, 2-EISA NetFlx-3 10/100, PCI Smart/2-Array Controller Raid 0, 1 SCSI Disk, MaxReceive Buffers is 100, HTTPD cache is on, DNS Lookups for HTTPD is on, and 100Mb Network. The graph of the base system run is displayed in Graph 1. Base Run- Run #1 1100 1000 900 800 TPM 700 600 500 400 300 200 100 0 1 12 24 32 36 48 Number of Virtual Clients 60 72 Graph 1: Base Run for HTTP and FTP Transactions The jump from 300-TPM on 1 virtual client to 590-TPM on 12 virtual clients is expected as the load increases on the system. The slight dip to 470-TPM on 32 virtual clients is due to garbage collection and cleanup of the firewall system and is noticed throughout all of the test runs. From 36 virtual clients through 72 virtual clients the TPM is maintained steadily between 560-TPM's through the 570-TPM's. The average number of failures for all runs was under 1%. Memory Run #2 and #3 increase memory from the base system up to 128 MB and 256 MB respectively. Raptor's Eagle NT 3.06 product is made to run with no less than 32 MB with a static virtual memory size of 200 MB. As displayed in Graph 2, increasing the memory of the firewall system only adds performance value when the load gets heavier. Tests with 36 to 72 virtual clients show increases in performance up to 3%, where 12 to 32 virtual clients show decreases in performance by 1-2% for lighter loads, for both memory scenarios. The reason for this behavior is that the firewall only needs a set amount of memory to process connections. Once the connection is established for each virtual client, the process of moving data between the incoming connection and outgoing connection does not require the use of more memory. 15

  • 1
  • 2
  • 3
  • 4
  • 5
  • 6
  • 7
  • 8
  • 9
  • 10
  • 11
  • 12
  • 13
  • 14
  • 15
  • 16
  • 17
  • 18
  • 19
  • 20
  • 21
  • 22
  • 23
  • 24
  • 25
  • 26
  • 27
W
HITE
P
APER
(cont.)
15
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
278A/0497
subsection contains some test runs based on lower end ProLiant and ProSignia systems and was
run to show performance on lower-end systems.
Base System
The base system, test run 1, consists of the ProLiant 5000 system, 1Pentium Pro 200 MHz, 512K
cache processor, 64 MB RAM, 2-EISA NetFlx-3 10/100, PCI Smart/2-Array Controller Raid 0, 1
SCSI Disk, MaxReceive Buffers is 100, HTTPD cache is on, DNS Lookups for HTTPD is on,
and 100Mb Network.
The graph of the base system run is displayed in Graph 1.
Base Run- Run#1
0
100
200
300
400
500
600
700
800
900
1000
1100
1
12
24
32
36
48
60
72
Number of Virtual Clients
TPM
Graph 1:
Base Run for HTTP and FTP Transactions
The jump from 300-TPM on 1 virtual client to 590-TPM on 12 virtual clients is expected as the
load increases on the system. The slight dip to 470-TPM on 32 virtual clients is due to garbage
collection and cleanup of the firewall system and is noticed throughout all of the test runs. From
36 virtual clients through 72 virtual clients the TPM is maintained steadily between 560-TPM’s
through the 570-TPM’s.
The average number of failures for all runs was under 1%.
Memory
Run #2 and #3 increase memory from the base system up to 128 MB and 256 MB respectively.
Raptor’s Eagle NT 3.06 product is made to run with no less than 32 MB with a static virtual
memory size of 200 MB. As displayed in Graph 2,
increasing the memory of the firewall system
only adds performance value when the load gets heavier. Tests with 36 to 72 virtual clients show
increases in performance up to 3%, where 12 to 32 virtual clients show decreases in performance
by 1-2% for lighter loads, for both memory scenarios. The reason for this behavior is that the
firewall only needs a set amount of memory to process connections. Once the connection is
established for each virtual client, the process of moving data between the incoming connection
and outgoing connection does not require the use of more memory.