KEMBAR78
Compression Performance | PDF | Transmission Control Protocol | Computer Science
0% found this document useful (0 votes)
9 views10 pages

Compression Performance

The document presents the compression performance of the BIG-IP 6800 (v9.0.5) using various test configurations and parameters. It includes details on the test environment, CPU utilization versus throughput, and the impact of different gzip levels on performance. Results indicate varying levels of CPU utilization and bytes saved through compression, emphasizing the importance of file type and size in compression efficiency.

Uploaded by

other.yeung
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
9 views10 pages

Compression Performance

The document presents the compression performance of the BIG-IP 6800 (v9.0.5) using various test configurations and parameters. It includes details on the test environment, CPU utilization versus throughput, and the impact of different gzip levels on performance. Results indicate varying levels of CPU utilization and bytes saved through compression, emphasizing the importance of file type and size in compression efficiency.

Uploaded by

other.yeung
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 10

Compression

Compression Performance
BIG-IP 6800 (v9.0.5)
4/7/2005
Compression Performance: BIG-IP 6800 (V9.0.5)
Test Environment

•Ixia 1600T chassis populated with 8 ALM1000T8


application load modules
•IxOS v3.80 and IxLoad v1.10 SP2 used to generate
test traffic

•External client and


internal server traffic
configuration: 4
Gigabit aggregated
and tagged twisted
pair

2
Compression Performance: BIG-IP 6800 (V9.0.5)
BIG-IP Configuration
From bigip.conf: From bigip_base.conf:

node * monitor icmp mgmt 5.5.5.81 {


profile http compression { netmask 255.255.255.0
compress enable
}
compress buffer size 65536
compress gzip memory level 32k trunk trunk1 {
(gzip level varies
from 1-9) compress gzip window size 64k interface 1.1 1.15 1.16 1.2
compress gzip level 1 }
compress cpu saver disable vlan internal_31 {
}
tag 31
pool pool_8080 {
monitor all tcp trunks tagged trunk1
member 30.30.30.1:webcache }
member 30.30.30.2:webcache vlan external_41 {
member 30.30.30.3:webcache tag 41

trunks tagged trunk1
member 30.30.30.32:webcache
} }
virtual vs_http_compress { stp instance 0 {
destination 40.40.40.202:http vlan external_41
ip protocol tcp vlan internal_31
profile compression tcp
trunk trunk1 external path cost 20000 internal path cost 20000
pool pool_8080
} }
self 30.30.30.254 {
netmask 255.255.0.0
vlan internal_31
}
self 40.40.40.254 {
netmask 255.255.0.0
vlan external_41
} 3
Compression Performance: BIG-IP 6800 (V9.0.5)
Test Parameters: CPU Utilization vs. Throughput

• 32 Ixia ALM1000T8 client ports


• 32 Ixia ALM1000T8 server ports

• 1000 users / concurrent connections


• Microsoft IE 5.x browser emulation
• HTTP 1.1 / 10 transactions per connection
• 44KB HTML request
• Gzip level varied from 1-9

4
Compression Performance: BIG-IP 6800 (V9.0.5)
Test Results: CPU Utilization vs. Throughput (gzip levels 1-9)

CPU Utilization vs. Throughput


6800 - v905 build 59.7 - 4/7/2005
(requested file: 44k.htm )

100%

90%

80%

70%
gzip level 1
CPU Utilization (%)

60% gzip level 2


gzip level 3
50%
gzip level 4
40% gzip level 5

30% gzip level 6


gzip level 7
20%
gzip level 8

10% gzip level 9

0%
0.0 100.0 200.0 300.0 400.0 500.0 600.0
Server TX + Client TX (Precompressed Layer 7 Mbps)

5
Compression Performance: BIG-IP 6800 (V9.0.5)
Test Results: CPU Utilization vs. Throughput
(Hardware compression and gzip levels 1-9)

CPU Utilization vs. Throughput


6800 - v905 build 59.7 - 4/7/2005
(requested file: 44k.htm)

100%

90%

80%
gzip level 1
70% gzip level 2
CPU Utilization (%)

60% gzip level 3

gzip level 4
50%
gzip level 5

40% gzip level 6

gzip level 7
30%
gzip level 8
20% gzip level 9

10% Hardware Comp

0%
0.0 200.0 400.0 600.0 800.0 1000.0 1200.0 1400.0 1600.0 1800.0
Server TX + Client TX (Precompressed Layer 7 Mbps)

6
Compression Performance: BIG-IP 6800 (V9.0.5)
Test Results: Bytes saved by using compression*
Bytes Saved (Server Bytes Sent vs. Client Bytes Received)
6800 - v905 build 59.7 - 4/7/2005
(using 44k.htm)
1800.0

1581.6
1600.0

1400.0 Maximum Server TX:

Maximum Client RX:


1200.0
Layer 7 Mbps

1000.0

800.0
492.1
485.2

600.0
417.7

412.9

289.7

271.4

255.6
400.0

246.2

208.6

173.5
200.0
84.3

64.8

62.1

41.7

37.9

34.3

32.9

28.0

23.0
0.0
l1

l2

l3

l4

l5

l6

l7

l8

l9
n o

ve

ve

ve

ve

ve

ve

ve

ve

ve
si
es

le

le
le

le

le

le

le

le

le
pr

ip

ip

ip

ip

ip

ip

ip

ip

ip
om

gz

gz

gz

gz

gz

gz

gz

gz

gz
C
e
ar
dw
ar
H

Note: Unless interpreted properly, this data can be misleading. These tests were run using the “44k.htm” file
used in the tests shown on the previous slides. Results will vary based on the type, size and compressibility of
the file, gzip level and other factors. Use this chart for relative comparison between gzip levels only.
7
Compression Performance: BIG-IP 6800 (V9.0.5)
Test Parameters: Concurrency

• 32 Ixia ALM1000T8 client ports


• 32 Ixia ALM1000T8 server ports

• 1000 users / concurrent connections


• Microsoft IE 5.x browser emulation
• HTTP 1.1 / unlimited transactions per connection
• 16KB HTML request
• Ramp-up rate = 96 new connections per second
• Concurrent connections established = 50,000

8
Compression Performance: BIG-IP 6800 (V9.0.5)
Test Results: Concurrency
Compression Concurrency
6800 - v9.0.5 build 59.7 - 4/5/2005
(Requested file size = 16KB)

1600 25000

1400

20000

Response Time = Connect Time (ms) + TTLB (ms)


1200 Gzip Level 1 - Mbps
Hardware Compression - Mbps
Gzip Level 1 - Response Time
Precompressed Layer 7 Mbps

1000
Hardware Compression - Response Time 15000

800

10000
600

400

5000

200

0 0
0 5000 10000 15000 20000 25000 30000 35000 40000 45000 50000
Concurrent Connections

You might also like