Cisco Cisco UCS B230 M1 Blade Server Libro blanco
Cisco blade solution vs. HP blade solution: Full
enclosure power testing
enclosure power testing
TEST REPORT
MARCH 2010
KEY FINDINGS
The Cisco blade solution achieved up to 10.7%
more SPECjbb2005 bops/watt than the HP blade
solution. (See Figure 1.)
more SPECjbb2005 bops/watt than the HP blade
solution. (See Figure 1.)
The Cisco blade solution used 10.2% less power
per blade than the HP blade solution while
running Prime95 torture tests. (See Figure 2.)
per blade than the HP blade solution while
running Prime95 torture tests. (See Figure 2.)
The Cisco blade solution used 3.3% less power
per blade while idle than the HP blade solution.
(See Figure 3.)
per blade while idle than the HP blade solution.
(See Figure 3.)
When including a pair of redundant fabric
interconnects, the Cisco blade solution still would
achieve better power results than the HP blade
solution at sixteen blades (two enclosures for the
Cisco blade solution). With additional enclosures,
Cisco’s power advantage would increase. (See
Figure 4.)
interconnects, the Cisco blade solution still would
achieve better power results than the HP blade
solution at sixteen blades (two enclosures for the
Cisco blade solution). With additional enclosures,
Cisco’s power advantage would increase. (See
Figure 4.)
Executive summary
Cisco Systems
®
, Inc. (Cisco) commissioned
Principled Technologies (PT) to compare the power
per blade and performance per watt of the Cisco
UCS 5108 chassis and HP BladeSystem c7000
enclosures, fully populated with similarly configured
blade servers:
per blade and performance per watt of the Cisco
UCS 5108 chassis and HP BladeSystem c7000
enclosures, fully populated with similarly configured
blade servers:
Eight Cisco UCS B200 M1 blade servers in
a Cisco UCS 5108 Blade Server Chassis
(Cisco blade solution)
a Cisco UCS 5108 Blade Server Chassis
(Cisco blade solution)
Sixteen HP ProLiant BL460c G6 servers in
an HP BladeSystem c7000 Enclosure (HP
blade solution)
an HP BladeSystem c7000 Enclosure (HP
blade solution)
To compute power per blade, we divided the chassis
power by the number of blades for each blade
solution.
To test the power usage and performance per watt,
we used the Prime95 and SPECjbb
®
2005 test tools.
For more information about Prime95 and SPECjbb2005, see the Workload section below. For detailed data on
Prime95 results, as well as our SPECjbb2005 performance data, see the Test results section.
As Figure 1 shows, the Cisco blade solution achieved up to 10.7 percent more SPECjbb205 bops/watt than the
HP blade solution. The Cisco blade solution delivered a SPECjbb2005 result of 1,751.0 overall SPECjbb2005
bops/watt compared to the HP blade solution, which yielded 1,582.4 overall SPECjbb2005 bops/watt. We
calculated our score by taking the sum of the SPECjbb2005 bops on each blade divided by the average power
reading during the run. Higher numbers are better.
Prime95 results, as well as our SPECjbb2005 performance data, see the Test results section.
As Figure 1 shows, the Cisco blade solution achieved up to 10.7 percent more SPECjbb205 bops/watt than the
HP blade solution. The Cisco blade solution delivered a SPECjbb2005 result of 1,751.0 overall SPECjbb2005
bops/watt compared to the HP blade solution, which yielded 1,582.4 overall SPECjbb2005 bops/watt. We
calculated our score by taking the sum of the SPECjbb2005 bops on each blade divided by the average power
reading during the run. Higher numbers are better.
0
200
400
600
800
1,000
1,200
1,400
1,600
1,800
2,000
B
o
p
s
p
e
r
w
a
tt
SPECjbb2005 benchmark: Performance per watt
(higher numbers are better)
Cisco blade
solution
solution
HP blade
solution
solution
Figure 1: SPECjbb2005 bops/watt results for the Cisco and HP blade solutions. Higher numbers are better.
Cisco is
10.7% better