Overview

This chapter contains the following sections:

About the Optimized Multi-Node Setup

In Cisco UCS Director versions prior to Release 6.7.4.x, the multi-node setup included the following nodes:

  • One primary node

  • One or more service nodes

  • One monitoring database

  • One inventory database

Similar to 6.7.4.x release, 6.8.0.0 also support same capabilities and scale with the following nodes:

  • One database node

  • One primary node


Note

After upgrading to release 6.7.4.x, since the multi-node configuration requires only 2 VMs, you can claim the freed up VMs.


Minimum System Requirements for Optimized Multi-Node Setup

System Requirements for the Primary Node

Number of VMs

vCPU Allocation

Memory Allocation (GB)

Memory Reservation (GB)

Disk Size (GB)

Inframgr Memory Allocation (GB)

1 - 5000

4

16

16

100

8

5001 - 10000

4

22

22

100

12

10001 - 15000

4

28

28

100

12

15001 - 20000

4

34

34

100

16

20001 - 25000

8

40

40

100

16

25001 - 30000

8

46

46

100

24

30001 - 35000

8

52

52

100

24

35001 - 40000

8

58

58

100

28

40001 - 45000

8

64

64

100

28

45001 - 50000

8

64

64

100

32

You can configure the Inframgr memory allocation in the /opt/infra/bin/inframgr.env file.

System Requirements for the Database Node

Number of VMs

vCPU Allocation

Memory Allocation (GB)

Memory Reservation (GB)

Disk Read I/O Bandwidth (MBps)

Disk Write I/O Bandwidth (MBps)

Disk Size (GB)

Database InnoDB Buffer Pool (GB)

1 - 5000

4

12

12

4

4

100

8

5001 - 10000

4

16

16

6

6

100

12

10001 - 15000

4

28

28

8

8

100

24

15001 - 20000

4

40

40

10

10

200

36

20001 - 25000

8

52

52

12

12

200

48

25001 - 30000

8

64

64

14

14

200

60

30001 - 35000

8

76

76

16

16

300

72

35001 - 40000

16

90

90

18

18

600

84

40001 - 45000

16

90

90

20

20

600

84

45001 - 50000

16

90

90

22

22

600

84

You can configure the Database InnoDB Buffer Pool parameter in the /etc/my.cnf file.


Note

To determine the currently configured disk read I/O bandwidth and disk write I/O bandwidth, use the Collect Diagnostics option from the Cisco UCS Director Shell Admin menu.


Database Parameters

VMs

Thread Cache Size

Maximum Connections

innodb lock wait timeout

Query Cache Size (MB)

Maximum Connection Errors

Connection Timeout

innodb read I/O Threads

innodb write I/O Threads

1 - 5000

1000

1000

100

128

10000

20

64

64

5001 - 10000

1000

1000

100

128

10000

20

64

64

10001 - 15000

1000

1000

100

128

10000

20

64

64

15001 - 20000

1000

1000

100

128

10000

20

64

64

20001 - 25000

2000

2000

100

128

10000

20

64

64

25001 - 30000

2000

2000

100

128

10000

20

64

64

30001 - 35000

4000

2000

100

128

10000

20

64

64

35001 - 40000

4000

4000

100

128

10000

20

64

64

40001 - 45000

4000

4000

100

128

10000

20

64

64

45001 - 50000

4000

4000

100

128

10000

20

64

64

Configure these parameters in the /etc/my.cnf file.

Guidelines and Limitations for Optimized Multi-Node Setup

Before you configure an optimized multi-node setup, review the following guidelines:

  • Plan the locations and IP addresses of your nodes carefully. You cannot change the types of any nodes. For example, you cannot reconfigure a database node as a primary node or a primary node as a database node.

  • Install licenses only on the primary node.

  • After you configure the nodes, the list of operations available in the shelladmin changes for the database node and the primary node.

  • If you modify your standalone configuration to a multi-node setup, you cannot revert to the standalone configuration unless you took a snapshot of the standalone configuration.

  • Connector packs are installed only on the primary node, while Cisco UCS Director patch releases are installed on the database node. As a result, you may notice a discrepancy of software versions between the primary node and the database node.

Best Practices for an Optimized Multi-Node Setup

Before you configure a multi-node setup for Cisco UCS Director, consider the following best practices:

  • To maximize output and minimize network latency, we recommend that the primary node and the database node reside on the same host.

  • Network latency (average RTT) between the primary or service node and the physical, virtual compute, storage, and network infrastructures should be minimized. A lower average RTT results in increased overall performance.

  • You can reserve more CPU cycles (MHz) and memory than recommended for better performance at system load.

    See System Requirements for the Primary Node and System Requirements for the Database Node.

  • You must configure passwordless authentication between the application node and the database node to:

    • Use the backup and restore feature available in Cisco Intersight on claimed UCS Director instances.

    • Enforce default password reset capability for SSH root and shelladmin users.

      You are prompted to reset the default SSH root user and shelladmin user passwords before logging into the Cisco UCS Director administrator interface. You will be prompted to reset these passwords only if you have not reset the passwords prior to upgrading to release 6.7(4.0). In an optimized multi-node environment, you must reset the password for these user accounts on the application node and the database node.

    See Setting Up Passwordless Authentication.