Cisco Prime Optical Installation Guide, 9.3

System Requirements

This chapter describes what is required to install Cisco Prime Optical 9.3. It contains the following sections:

•Prime Optical Server Requirements

•Prime Optical Client Requirements

•Oracle Licensing for Prime Optical

•Overview of Sudo Commands

Note Although Cisco makes every attempt to ensure the availability of third-party hardware and software platforms specified for Prime Optical, Cisco reserves the right to change or modify system requirements due to third-party vendor product availability or changes that are beyond Cisco's control.

1.1  Prime Optical Server Requirements

The Prime Optical server runs on any of the platforms listed in Table 1-1.

Note•The Prime Optical server must run on a dedicated workstation. Any application that is not explicitly listed in this chapter as being required or supported by Prime Optical cannot be installed on the dedicated Prime Optical server workstation.

•Prime Optical is not validated against required file systems (/oracle, /db01, /db02, /db03, /db04, /db05) of the Network File System (NFS) type.

•Prime Optical is supported on Oracle VM servers for SPARC (previously called Sun logical domains [LDOMs]). The virtual server (logical domain) must comply with processor, memory, and disk requirements based on system size.

•Prime Optical does not support local Solaris zones.

•When Solaris 10 is installed, its Windows manager, Java Desktop System (JDS), is also installed. The first time you start Solaris 10, you are prompted with a message to choose Common Desktop Environment (the legacy desktop system) or JDS.

•Before you install the Prime Optical server and the Oracle 11g database on your Solaris or Linux server, verify the following:

–The ping command is included in your path environment variable.

–You have identified nonroot users and related UNIX groups to run Prime Optical UNIX commands. See Overview of Sudo Commands.

Table 1-2 lists the items required to install the Prime Optical server.

1.1.1  Server Specifications

The following tables show recommended hardware specifications for installing the Prime Optical server, and the resulting maximum number of NEs the server manages for each configuration.

If the Prime Optical server and Oracle database are installed on separate workstations, both workstations must meet the hardware requirements shown in the following tables.

Note It is possible to use and configure disk arrays for database storage with any type of RAID layout. You can do so only if the disk throughput is equal to or higher than the throughput calculated for internal disks.

Cisco tests the simulated network configurations listed in Table 1-5. Your setup and performance might vary depending on the size of your network and the usage pattern of management tasks.

The following list describes the server configuration parameters and the effect of changes in each parameter on the maximum number of NEs the server can manage:

•Network Size—If the network size increases, more resources are reserved for the higher number of NEs to be managed.

•RAM—The server can manage more NEs as the RAM increases. The server can manage fewer NEs as the RAM decreases.

•Number of Network Partitions—For each network partition, the server reserves resources for the higher number of NEs to be managed. The server can manage more NEs as the number of network partitions increases. The server can manage fewer NEs as the number of network partitions decreases.

1.1.2  Disk Space and Partition Specifications

The following tables show disk space and partition requirements for optical NEs based on network size and PM data collection status.

•The minimum disk space is the total space required for /, swap, /cisco, /oracle, /db01, /db02, /db03, /db04, and /db05. (The minimum disk space does not include /ctm_backup or the optional partitions /db01_rd and /db02_rd.)

•The maximum disk space is the total space required for /, swap, /cisco, /oracle, /db01, /db02, /db03, /db04, /db05, /db01_rd, /db02_rd, and /ctm_backup.

Note•It is recommended that you use the partition sizes described in this section. Prime Optical performance degrades if you do not use the recommended partition sizes.

•To improve database performance and maximize data input/output, configure a dedicated disk array with the following setup:

–Dynamic multipathing (DMP) to parallelize the file system access.

–Raid 0 (striping) to optimize the disk input data.

•Prime Optical is a high-transaction application, which causes high I/O throughput for Oracle redo logs. It is recommended that you reserve dedicated system controllers for file systems—including the optional /db01_rd and /db02_rd file systems—that manage redo logs.

Caution For large and high-end networks, if you do not follow the preceding disk striping recommendations, the I/O might create a critical bottleneck and cause an unexpected Prime Optical shutdown.

Note the following PM assumptions for optical NEs:

•In a small network, PM data collection assumes 30 days of storage for PM data collected across 200 NEs, assuming an average of 200 interfaces per NE, up to a maximum of 40,000 interfaces (includes logical and physical interfaces).

•In a medium network, PM data collection assumes 30 days of storage for PM data collected across 400 NEs, assuming an average of 100 interfaces per NE, up to a maximum of 40,000 interfaces (includes logical and physical interfaces).

•In a large network, PM data collection assumes 30 days of storage for PM data collected across 2000 NEs, assuming an average of 100 interfaces per NE, up to a maximum of 200,000 interfaces (includes logical and physical interfaces).

•In a high-end network, PM data collection assumes 30 days of storage for PM data collected across 5000 NEs, assuming an average of 100 interfaces per NE, up to a maximum of 500,000 interfaces (includes logical and physical interfaces).

1.1.2.1  Understanding the ctm_backup Directory

The ctm_backup directory is a repository used by the oracle user to back up the following main categories of information:

•Prime Optical database backup and export dump

•Configuration files

•Database ARCHIVELOG files

The oracle user must have read/write permissions or the database backup will fail. The database should be blocked if the ARCHIVELOG files cannot be moved to the /ctm_backup directory.

It is recommended that the /ctm_backup size be equivalent to the sum of the single database partitions (that is, the sum of /oracle, /db01, /db02, /db03, /db04, /db05, and—if configured—/db01_rd and /db02_rd).

Note•The final /ctm_backup size is also related to the ARCHIVELOG files.

•The /ctm_backup directory can be a symbolic link to a user-defined directory and must have read/write permissions.

1.2  Prime Optical Client Requirements

To install the Prime Optical client, you must have a Linux or Solaris workstation or Windows PC. Memory, CPU, and disk requirements are reported in Table 1-11.

The delta RAM and delta CPU must be counted for each additional launched client.

It is strongly recommended that you install the Prime Optical client on a workstation separate from the Prime Optical server. Installing the Prime Optical client and server on the same workstation consumes server resources and could degrade performance.

Table 1-12 shows the maximum number of simultaneous Prime Optical client sessions.

1.2.1  Using Remote Application Software with the Prime Optical Client

Client launch and operation are supported by the following remote application software:

•Windows Server 2003 Enterprise Edition with Terminal Services

•Citrix Presentation Server 4.0

•Secure Global Desktop Enterprise Edition 4.0 (previously known as Tarantella Enterprise)

Note If you are using Secure Global Desktop, enable full-duplex autodetection on the GUI server interface to prevent performance slowdown.

The hardware requirements for the remote application depend on the number of clients that the system must export, calculated with the following formula:

Target RAM = base RAM + (delta RAM x number of clients)

Target CPU = base CPU + (delta CPU x number of clients)

where:

•base RAM—Amount of RAM required by the remote application software.

•base CPU—Amount of CPU required by the remote application software.

•delta RAM—Amount of RAM required for each additional client.

•delta CPU—Amount of CPU required for each additional client.

1.2.1.1  Citrix and Windows Server 2003 Terminal Services

The following figure shows the environment for a remote Citrix Presentation Server or Windows Server 2003 Terminal Server. In this example, the GUI application server and the presentation server reside on the same workstation.

Figure 1-1 Remote Citrix or Windows Server 2003 Terminal Server Environment

In this example, the hardware requirements for 10 clients are:

•CPU = base CPU + (240 MHz x 10)

•RAM = base RAM + (512 MB x 10)

Note For the base RAM and base CPU requirements, refer to the documentation for:

•Windows Server 2003 Enterprise Edition with Terminal Services

•Citrix Presentation Server 4.0

1.2.1.2  Secure Global Desktop Enterprise Edition 4.0

The following figure shows the environment for a remote client via the Secure Global Desktop server. In this example, the GUI application server and the Secure Global Desktop server reside on different workstations. The application server is where Prime Optical clients run.

Figure 1-2 Remote Secure Global Desktop Environment

The following example shows requirements for a remote Secure Global Desktop server on a Solaris workstation and a GUI application server on a Windows PC.

In this example, the hardware requirements for 10 clients are:

•Application server CPU = base CPU + (240 MHz x 10)

•Application server RAM = base RAM + (512 MB x 10)

Note•For the Secure Global Desktop server CPU and RAM requirements, refer to the Secure Global Desktop documentation.

•For the application server base RAM and base CPU requirements, refer to the documentation for Windows Server 2003 Enterprise Edition with Terminal Services.

The following example shows requirements for a remote Secure Global Desktop server on a Solaris workstation and a GUI application server on another Solaris workstation.

In this example, the hardware requirements for 10 clients are:

•Application server CPU = base CPU + (240 MHz x 10)

•Application server RAM = base RAM + (512 MB x 10)

Note•For the Secure Global Desktop server CPU and RAM requirements, refer to the Secure Global Desktop documentation.

•For the application server base RAM and base CPU requirements, refer to the documentation for Windows Server 2003 Enterprise Edition with Terminal Services.

Using the Prime Optical Client in a Sun Ray Environment

The Prime Optical client is supported in an Oracle Sun Ray 2 environment. The setup uses thin terminal devices and Oracle Sun Ray 2 virtual display clients that are network-connected to a server running Sun Ray server software.

A virtual desktop runs on the Sun Ray server, while the related display is exported remotely to the associated Sun Ray 2 virtual display client.

•For Prime Optical client requirements, see Table 1-11.

•For Oracle Sun Ray 2 virtual display client requirements and network and software configurations, see:

–http://www.oracle.com/us/products/servers-storage/desktop-workstations/030726.htm

–http://www.oracle.com/us/products/servers-storage/desktop-workstations/036088.pdf

1.2.2  Java Heap Sizes

The Prime Optical client startup script provides memory allocation and identifies the maximum heap allocation for the client Java Virtual Machine (JVM) process. The Prime Optical client launches with a maximum heap size of 512 MB.

1.3  Oracle Licensing for Prime Optical

This section explains how to calculate the total number of Oracle Named User Plus licenses required for your Prime Optical server and client installations. This section describes with examples the Oracle Enterprise Edition database edition.

Note See the Oracle website for detailed information about Oracle licensing definitions and requirements.

Oracle's technology products, including the Oracle database required for use with Prime Optical, are licensed using one of two possible metrics. Which metric to use normally depends on which will result in a lower price for the database. If for some reason the number of database users cannot be counted, the Processor metric must be used. However, in a Prime Optical environment, database users normally can be counted:

•Processor—This metric is defined as the number of processors on the server on which the Oracle database is installed or running. This option must be used in environments in which software users cannot easily be identified or counted (not normally the case in a Prime Optical environment).

Note For the purpose of counting the number of processors, a multicore chip with n cores is determined by multiplying n cores by a core processor licensing factor of 0.75. All cores on all multicore chips for each licensed program for each core processor licensing factor listed above are aggregated, before multiplying by the appropriate core processor licensing factor. All fractions of a number are rounded up to the next whole number.

•Named User Plus—This metric is used in environments in which users can be identified and counted. Named User Plus includes both human-operated and automated devices. All human-operated and automated devices that access the program must be licensed. A Named User Plus license may access the program on any instances on which it is deployed, provided that the minimum license requirement on each server is met.

In the context of a Prime Optical environment, human-operated device means any device operated by a user who has direct or indirect access to Prime Optical. Direct access is gained through a user account on Prime Optical that allows access through the Prime Optical client GUI. Indirect access is possible through a user account on a higher-layer OSS, which in turn communicates with Prime Optical through Prime Optical GateWay/CORBA. Automated users include the NEs managed by Prime Optical and the Prime Optical server itself.

Oracle Enterprise Edition requires a minimum of 25 Named User Plus licenses per processor or the total number of actual users, whichever number is higher.

The following figure shows an example Prime Optical environment to illustrate identifying the human and automated database users that must be counted.

Figure 1-3 Example Prime Optical Environment

In the example shown in Figure 1-3 there are two independent Prime Optical servers and Oracle database instances. There are four NEs, one of which is managed by both Prime Optical servers. There is one direct Prime Optical user and one indirect user. So in this example the total number of Oracle named users is as follows:

Prime Optical servers: 2*

NEs: 4**

Prime Optical users: 1**

Higher-layer OSS users: 1

Total Named User Plus: 8 = 2 + 4 + 1 + 1

*Because of the self-monitor feature of Prime Optical, a Prime Optical server itself is considered an automated user of the database and is therefore counted.

**A Named User Plus license entitles the user to access Oracle on any instances where it is deployed. So if a user has access to multiple Prime Optical servers, only a single license is needed per user. Also, if an NE is managed by multiple Prime Optical servers, only a single license is needed per NE.

This example explains how to count the named users, but the number of named users required is the larger of either 1) the actual count or 2) the required minimum.

1.3.1  Oracle Enterprise Edition

Oracle Database Enterprise Edition (EE) provides improved scalability performance. The two major features in this release are:

•The Oracle partitioning option

•An increased maximum number of supported CPUs

Note Oracle EE requires a separate license.

1.3.1.1  Prime Optical with Oracle EE (Example A)

A service provider has 800 NEs and a data center with 10 Prime Optical client workstations. 100 employees in the data center are authorized to use the Prime Optical client. Some of the employees in the data center share the same Prime Optical account (username/password). Prime Optical is running on a Sun V880 with 8 processors, with only one core and 32 GB of RAM.

•Named User Plus: 1) Minimum = 8 processors x 25 users/processor = 200 or 2) Count = 800 licenses for NEs + 100 licenses for data center personnel + 1 Prime Optical server = 901

Result: 901 Named User Plus licenses are required

•Processor: 8 processor licenses

If the processors are UltraSPARC IV and IV+ (dual core), the calculation for processor licensing is:

•Processor: (8 x 2) x 0.75 = 12 processor licenses

1.3.1.2  Prime Optical with Oracle EE (Example B)

The same service provider as in the previous example decides to enable Prime Optical GateWay/CORBA and connect to a higher-layer OSS that handles inventory management. There are 20 employees in the data center authorized to access the inventory system; 5 of them are also Prime Optical users (that is, they are a subset of the 100 Prime Optical users identified in the previous example). In this case, the total number of human users is:

95 Prime Optical-only users + 15 inventory system-only users + 5 Prime Optical/inventory system users, or 115 human named users.

As a result of adding the OSS and related users, the total number of named users required has increased from 901 to 916.

1.4  Overview of Sudo Commands

Sudo software (freeware) version 1.6.6 is bundled with the Prime Optical software. The sudo software enables nonroot UNIX users to run the following UNIX commands:

•ctms-start

•ctms-abort

•ctms-stop

•ctms-stop-service

•showctm

•getinfo.sh

•prune_auditlog.sh

•prune_errlog.sh

•prune_audittrail.sh

•prune_fm.sh

•prune_pm.sh

•prune_ne.sh

•prune_server_monitor.sh

•prune_admin_job_table.sh

•prune_ne_ipaddress.sh

During the Prime Optical server installation, the setup program prompts you to specify the name of the UNIX group to which you want to assign administrator privileges. By default, this group is set to the root group. If you specify a group other than root, the setup program verifies that the UNIX group exists on the system and adds entries to the /etc/sudoers file. Entries in this file reflect the commands that the specified UNIX group can run by using the sudo command.

The following entries in the /etc/sudoers file reflect the commands that can be run as nonroot:

%CTM_UNIX_group hostname=(root) NOPASSWD: /usr/bin/prune_admin_job_table.sh

%CTM_UNIX_group hostname=(root) NOPASSWD: 
/opt/CiscoTransportManagerServer/bin/prune_admin_job_table.sh

%CTM_UNIX_group hostname=(root) NOPASSWD: /usr/bin/prune_server_monitor.sh

%CTM_UNIX_group hostname=(root) NOPASSWD: 
/opt/CiscoTransportManagerServer/bin/prune_server_monitor.sh

%CTM_UNIX_group hostname=(root) NOPASSWD: /usr/bin/prune_ne.sh

%CTM_UNIX_group hostname=(root) NOPASSWD: /opt/CiscoTransportManagerServer/bin/prune_ne.sh

%CTM_UNIX_group hostname=(root) NOPASSWD: /usr/bin/prune_pm.sh

%CTM_UNIX_group hostname=(root) NOPASSWD: /opt/CiscoTransportManagerServer/bin/prune_pm.sh

%CTM_UNIX_group hostname=(root) NOPASSWD: /usr/bin/prune_fm.sh

%CTM_UNIX_group hostname=(root) NOPASSWD: /opt/CiscoTransportManagerServer/bin/prune_fm.sh

%CTM_UNIX_group hostname=(root) NOPASSWD: /usr/bin/prune_audittrail.sh

%CTM_UNIX_group hostname=(root) NOPASSWD: 
/opt/CiscoTransportManagerServer/bin/prune_audittrail.

%CTM_UNIX_group hostname=(root) NOPASSWD: /usr/bin/prune_errlog.sh

%CTM_UNIX_group hostname=(root) NOPASSWD: 
/opt/CiscoTransportManagerServer/bin/prune_errlog.

%CTM_UNIX_group hostname=(root) NOPASSWD: /usr/bin/prune_auditlog.sh

%CTM_UNIX_group hostname=(root) NOPASSWD: 
/opt/CiscoTransportManagerServer/bin/prune_auditlog.

%CTM_UNIX_group hostname=(root) NOPASSWD: /usr/bin/getinfo.sh

%CTM_UNIX_group hostname=(root) NOPASSWD: /opt/CiscoTransportManagerServer/bin/getinfo.sh

%CTM_UNIX_group hostname=(root) NOPASSWD: /usr/bin/showctm

%CTM_UNIX_group hostname=(root) NOPASSWD: /opt/CiscoTransportManagerServer/bin/showctm

%CTM_UNIX_group hostname=(root) NOPASSWD: /usr/bin/ctms-stop-service

%CTM_UNIX_group hostname=(root) NOPASSWD: 
/opt/CiscoTransportManagerServer/bin/ctms-stop-service

%CTM_UNIX_group hostname=(root) NOPASSWD: /usr/bin/ctms-stop

%CTM_UNIX_group hostname=(root) NOPASSWD: /opt/CiscoTransportManagerServer/bin/ctms-stop

%CTM_UNIX_group hostname=(root) NOPASSWD: /usr/bin/ctms-abort

%CTM_UNIX_group hostname=(root) NOPASSWD: /opt/CiscoTransportManagerServer/bin/ctms-abort

%CTM_UNIX_group hostname=(root) NOPASSWD: /usr/bin/ctms-start

%CTM_UNIX_group hostname=(root) NOPASSWD: /opt/CiscoTransportManagerServer/bin/ctms-start

1.4.1  Using the ctms-start Command

A complete set of administrative command scripts is added to the application during installation. One command automatically starts the Prime Optical server processes every time the server is started. The server processes can also be started or stopped manually as necessary; the scripts are located in the /opt/CiscoTransportManagerServer/bin directory.

The ctms-start command sets the appropriate environment variables and starts the Prime Optical server. The amount of time it takes for the Prime Optical server to start varies based on the number of NEs in the configuration and the size of the database. Use ctms-start only when the Prime Optical server has stopped.

Step 1 Log into the Prime Optical server workstation as the root user.

Step 2 On the command line, enter the following command:

ctms-start

Note It can take from 0 to 5 minutes for the server processes to start after the ctms-start command has finished execution. This is because NE services and gateway services (if enabled) are still initializing for all of the NEs that are deployed. Wait 5 minutes after entering the ctms-start command; then, enter the showctm command. The NE service corresponding to all the deployed NEs should have started.

1.4.2  Using the ctms-stop Command

The ctms-stop command stops the Prime Optical server gracefully. The stop procedure shuts down the server and cleans all memory and connections. The overall process takes approximately 5 minutes.

Step 1 Log into the Prime Optical server workstation as the root user.

Step 2 On the command line, enter the following command:

ctms-stop

1.4.3  Using the ctms-abort Command

The ctms-abort command kills all of the running processes and stops the Prime Optical server. The overall process takes no longer than 2 to 3 minutes.

Step 1 Log into the Prime Optical server workstation as the root user.

Step 2 On the command line, enter the following command:

ctms-abort

1.4.4  Using the showctm Command

The showctm command provides Prime Optical version and process information.

Step 1 Log into the Prime Optical server workstation as the root user.

Step 2 On the command line, enter the following command:

showctm

The following is an example of the output of the showctm command on Solaris, where all of the attributes (except for the process names) are variable:

Cisco Prime Optical Server Version: 9.3 Build: build-number

-------------------------------------------------------------------------------------

USER   PID   %CPU       %MEM          START         TIME  PROCESS

-------------------------------------------------------------------------------------

root  25465  0.9   1.1910216706960    May_11    11342:55  CTM Server

root  25442  0.0   0.0166920          May_11        0:00  CTM Server

root  26868  0.1   0.2447368127760    May_11      173:47  SnmpTrapService

root  25550  0.1   0.3229432174840    May_11      238:23  SMService

root  26850  0.0   0.3215680167824    May_11       62:37  OrchestrationService

root  27986  0.8   6.750663204384832  May_11     3074:57  CTC-based SONET NetworkService-11

root  27844  0.4   6.850544964429952  May_11     4167:30  CTC-based SONET NetworkService-8

root  27583  0.4   6.750682564386936  May_11     4180:07  CTC-based SONET NetworkService-3

root  27736  0.3   6.850703204454040  May_11     3264:30  CTC-based SONET NetworkService-6

root  27632  0.3   6.950674724527640  May_11     4233:57  CTC-based SONET NetworkService-4

root  27938  0.2   6.950558084472240  May_11     4112:43  CTC-based SONET NetworkService-10

root  27795  0.1   6.850792004414888  May_11     4023:46  CTC-based SONET NetworkService-7

root  27534  0.1   7.050713364553928  May_11     4381:39  CTC-based SONET NetworkService-2

root  27484  0.2   1.41142320912192   May_11     2369:55  ONS15454/ONS15327/ONS15310CL/ONS15310MA PMService-1

root  23119  0.0   1.1839696725936    May_16      420:00  ONS15454/ONS15327/ONS15310CL/ONS15310MA PMService-9

root  27194  0.3   1.3976376860808    May_11     2989:55  CORBAGWService

root  25456  0.0   0.0                May_11        2:41  Apache Web Server

-------------------------------------------------------------------------------------

The following is an example of the output of the showctm command on Linux:

USER     PID    %CPU    %MEM     VSZ     RSS     TTY    STAT   START    TIME COMMAND

root    5751     0.0     0.3 1726440   88840       ?      Sl   May11    0:29 JBOSS JMS

root    5963     0.0     0.2 1227500   66752       ?      Sl   May11    0:03 NameService

root    6151     0.0     0.0  126960    4232       ?      Ss   May11    0:00 CTM Server

root    6166     0.0     0.0  120236    2424       ?      Ss   May11    0:00 Apache

root    6179     0.0     0.0 1711156   17592       ?      Sl   May11    0:00 CTM Server

root    6398     0.0     0.4  456656  101584       ?      Sl   May11    0:12 SMService

root    7248     0.0     0.4  470604  122928       ?      Sl   May11    0:11 Orchestration Service

root    7300     0.0     0.8 1235876  197860       ?      Sl   May11    0:10 SnmpTrapService

1.4.5  Using the ctms-stop-service Command

The ctms-stop-service command kills the service and starts a new instance of the service automatically.

Caution Use the ctms-stop-service command with extreme caution, because it terminates the service ungracefully. Before using this command, try to stop the service from the Control Panel window.

Step 1 Log into the Prime Optical server workstation as the root user.

Step 2 On the command line, enter one of the following commands to stop the Prime Optical process and automatically start a new service:

•SM service:

ctms-stop-service SMService 

•NE/PM services:

ctms-stop-serviceservice-ID-number

•Prime Optical GateWay/CORBA service:

ctms-stop-service -1

•SNMP trap service:

ctms-stop-service -2