Cisco Cisco UCS B460 M4 Blade Server 백서
© 2016 Cisco and/or its affiliates. All rights reserved. This document is Cisco Public information.
Page 7
Cisco UCS and Dell PowerEdge M1000e: A Comparison
June 2016
the same chassis. The worst-case scenario is three network hops required to
move data between servers in different chassis. With PowerEdge M1000e, traffic
between each chassis must first travel through the local Dell switch modules,
through an external ToR switch, and back through the second chassis’ Dell switch
module, resulting in three network hops and greater latency. Although Dell supports
pass-through modules, which can reduce the number of hops, this approach
greatly increases the cabling (and recabling) requirements associated with the
configuration. It also increases the number of ToR switch ports needed.
move data between servers in different chassis. With PowerEdge M1000e, traffic
between each chassis must first travel through the local Dell switch modules,
through an external ToR switch, and back through the second chassis’ Dell switch
module, resulting in three network hops and greater latency. Although Dell supports
pass-through modules, which can reduce the number of hops, this approach
greatly increases the cabling (and recabling) requirements associated with the
configuration. It also increases the number of ToR switch ports needed.
Actual east-west traffic tests of these identically configured systems show that:
• The PowerEdge M1000e with PowerEdge M I/O Aggregators has between 11
and 45 percent more latency than Cisco UCS. Cisco UCS demonstrated lower
latency than the PowerEdge M1000e with PowerEdge M I/O Aggregators for
every test case and every packet size (User Datagram Protocol [UDP], TCP, and
TCP round-trip times).
latency than the PowerEdge M1000e with PowerEdge M I/O Aggregators for
every test case and every packet size (User Datagram Protocol [UDP], TCP, and
TCP round-trip times).
• As packet sizes increased in each test, the PowerEdge M1000e disadvantage
also increased compared to Cisco UCS.
• Performance is almost identical for both single-chassis tests and multichassis
tests for Cisco UCS. With the Dell configuration, after traffic leaves the chassis,
latency increases dramatically.
latency increases dramatically.
Cisco remains the leader in application-specific integrated circuit (ASIC) and
network design optimization for the end-to-end network stack, enabling business
applications and virtual environments to perform better. The new, third-generation
Cisco UCS fabric interconnects allow Cisco UCS to have true 40-Gbps end-to-end
bandwidth, which Dell does not have, and can provide 360 Gbps of total bandwidth
to a chassis.
network design optimization for the end-to-end network stack, enabling business
applications and virtual environments to perform better. The new, third-generation
Cisco UCS fabric interconnects allow Cisco UCS to have true 40-Gbps end-to-end
bandwidth, which Dell does not have, and can provide 360 Gbps of total bandwidth
to a chassis.
UCS B200 M3
CONSOLE
UCS B200 M3
CONSOLE
UCS B200 M3
CONSOLE
UCS B200 M3
CONSOLE
UCS B200 M3
CONSOLE
UCS B200 M3
CONSOLE
UCS B200 M3
CONSOLE
UCS B200 M3
CONSOLE
1
2 3
4 5
6 7
8 9
10 11
12 13
14 15
16
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
28 29
30 31
32
1
2 3
4 5
6 7
8 9
10 11
12 13
14 15
16
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
28 29
30 31
32
UCS B200 M3
CONSOLE
UCS B200 M3
CONSOLE
UCS B200 M3
CONSOLE
UCS B200 M3
CONSOLE
UCS B200 M3
CONSOLE
UCS B200 M3
CONSOLE
UCS B200 M3
CONSOLE
UCS B200 M3
CONSOLE
Cisco Nexus® 5548 Switches
Cisco Nexus 5548 Switches
Switch
ASICs
Cisco UCS 6248UP Fabric Interconnects
Cisco UCS 5108 Blade Server Chassis
Cisco UCS 2208XP
Fabric Extenders
One Logical Chassis
Up to 160 Servers
Dell PowerEdge
M I/O Aggregators
Dell PowerEdge
M1000e Chassis
Two Physical and Logical
Chassis 16 Servers Each
A
X
Hop 1
Hop 1
Hop 2
Hop 3
Figure 3 Cisco UCS and Dell PowerEdge M1000e Traffic Flow Between Blade Servers in
Different Chassis
Different Chassis