Identifying Server Firmware Versions

• When managing servers using Intel® DCM, Server Firmware Outliers

users can see the firmware version of each
server
• That includes using the “firmware version
outlier” and “firmware version distribution”
features to show firmware versions on Server Firmware Outliers
servers of the same model
• This feature helps ensure that all servers are
running with the best performance,
reliability and security features possible

Department
Data Center Management
or Event Name
Solutions Intel Confidential 29
Intel® DCM
In Action
 Intel® DCM in Action
Real Customer Situations – Example 1

• Problem: Identify hotspots & cooling inefficiencies

Possible solutions:
Use perforated tiles Start fans in overheating chassis Safely increase set point temp.

Department
Data Center Management
or Event Name
Solutions Intel Confidential 31
 Intel® DCM in Action
Real Customer Situations – Example 2

• Problem: Identify underutilized servers

Possible solutions:
• EOL unused servers
• Temporarily shut down unused servers
• Move workloads and increase utilization
• Virtualize/consolidate underutilized servers

Department
Data Center Management
or Event Name
Solutions Intel Confidential 32
 Intel® DCM in Action
Real Customer Situations – Example 3

• Problem: Improve Data center Reliability

Possible solutions:
• Replace faulty components Connect lost PSU

• Migrate workloads away from faulty servers

• Don’t schedule workloads on faulty servers
• Perform firmware updates

Department
Data Center Management
or Event Name
Solutions Intel Confidential 33
 Intel® DCM in Action Power Consumption of all Nodes During Workload Execution
Real Customer Situations – Example 4 500

400

Power Consumption (W)

300

• Problem: Increase rack density 200

100

0.15:00

0:03:00

0:12:00

0:21:00
0:24:00
0:27:00

1:39:00
0.00:00

0:15:00

1:00:00

1:15:00

1:33:00
1:36:00

1:45:00

1:51:00
1:54:00
1:57:00
1:18:00

1:48:00
0.06:00
0.09:00

0.18:00
0.03:00

0:00:00

1:06:00
1:09:00
1:12:00

1:21:00
1:24:00

1:42:00
0.12:00

0.21:00
0.24:00
0.27:00

0:06:00
0:09:00

0:18:00

1:03:00

1:27:00
1:30:00
Possible solutions:
Time
Server - 192.168.121.101 Server - 192.168.121.102 Server - 192.168.121.103 Server - 192.168.121.104
Server - 192.168.121.105 Server - 192.168.121.106 Server - 192.168.121.106 Server - 192.168.121.107
Server - 192.168.121.108 Server - 192.168.121.109 Server - 192.168.121.110 Server - 192.168.121.111

• Perform rack level power 500

Power Consumption of all Nodes During Workload Execution

capping (in this case 6.25KW 400

instead of 8.6KW)

Power Consumption (W)

300

200

Add node-level min power

100

• 0

budget to minimize workload

0:03:00

0:12:00

0:21:00
0:24:00
0:27:00

1:15:00

1:45:00

1:51:00
1:54:00
1:36:00
1:39:00
0.15:00

1:33:00

1:57:00
0.00:00

0:15:00

1:00:00

1:06:00
1:09:00
0.06:00
0.09:00

1:18:00

1:48:00
0.18:00

0:00:00

1:03:00

1:12:00

1:21:00
1:24:00
1:27:00
1:30:00

1:42:00
0.03:00

0.12:00

0.21:00
0.24:00
0.27:00

0:06:00
0:09:00

0:18:00
impact (in this case 300W)
Time
Server - 192.168.121.101 Server - 192.168.121.102 Server - 192.168.121.103 Server - 192.168.121.104
Server - 192.168.121.105 Server - 192.168.121.106 Server - 192.168.121.106 Server - 192.168.121.107
Server - 192.168.121.108 Server - 192.168.121.109 Server - 192.168.121.110 Server - 192.168.121.111

Rack density increased from 16 to 22 nodes without impacting any single node more than the rest

Department
Data Center Management
or Event Name
Solutions Intel Confidential 34
Recent Whitepapers