Configuration and administration considerations, tasks, and use cases
This chapter describes the tasks that you must do to configure your Spectrum Control server to monitor, manage, and report on devices in your environment.
Additionally, this chapter provides administrative use cases for activities such as the following ones:
•Monitoring the Spectrum Control dashboard
•Cleaning up alerts, errors, and warnings in the dashboard
•Viewing inter-switch connections and their performance metrics
•Viewing connections for N_Port Virtualization (NPV) mode switches
4.1 Adding resources to Spectrum Control
One of your first actions after you complete a fresh installation is to add devices to Spectrum Control. The Add Device wizard in the Spectrum Control V5.2 GUI provides a simplified way for adding, discovering, and configuring devices.
The options to add a device depend on the type:
•To add storage systems, click Storage → Block Storage Systems, File Storage Systems, or Object Storage Systems. The types of storage systems that you can choose depends on the menu selection.
•To add fabrics and switches, click Network → Fabrics or Network → Switches.
•To add VMware Hypervisors or vCenters, click Servers → Hypervisors.
•To add Storage Resource agents, click Servers → Servers.
The following additional resources can be configured and added to Spectrum Control:
•To add agentless servers, click Servers → Servers.
•To add applications, click Groups → Applications.
•To add departments, click Groups → Departments.
|
Note: The configuration is performed in the web-based GUI for all device types. The stand-alone GUI from previous versions of Tivoli Storage Productivity Center no longer exists. For more information about the stand-alone GUI removal, see the Spectrum Control IBM Knowledge Center, found at:
|
The Add Device wizard has three main steps:
1. Select a resource.
2. Add and discover a device.
3. Configure a device.
Examples are provided in the subsequent sections for adding different device types. If you need additional details for configuring and administering the different device types as resources, see the IBM Spectrum Control IBM Knowledge Center, found at:
4.1.1 Adding IBM storage systems
Most storage systems from IBM are added by using the native API instead of a CIM agent. The type of information that you must provide to add the device to Spectrum Control depends on the device type.
This section describes the tasks to add a SAN Volume Controller to Spectrum Control. The tasks are similar for other IBM storage systems
Complete the following steps:
1. Check the support matrix for the required level of firmware before you begin adding a device at the following website:
From the support matrix page, click Spectrum Control 5.2.8+ in the column labeled Storage. A window with the list of supported storage systems and the supported firmware levels opens.
2. After verifying the required firmware levels, go to the Spectrum Control GUI to add the SAN Volume Controller. Click Storage Resources → Block Storage Systems. From the list of configured Block Storage Systems, click Add Storage System.
3. Select the type of storage system to add (see Figure 4-1). IBM devices are listed individually. “All others” should be used for other vendors.
Figure 4-1 Select the type of device for adding block storage systems
|
Tip: If you do not see the IBM device that you want, make sure that you selected the correct type of storage (block, file, or object) from the menu.
|
4. Clicking the SAN Volume Controller icon opens the Add Device wizard, as shown in Figure 4-2.
Figure 4-2 Specify connection details
5. Complete the connection information so that Spectrum Control can discover the devices that are available for this data source.
|
Note: Spectrum Control can add a SAN Volume Controller or Storwize V7000 server by using only a user ID and password instead of Secure Shell (SSH) keys for SAN Volume Controller Version 6.3 or later, which makes adding the devices simple. If a customer has strict rules about how often a password must be changed, using SSH keys might be the better option.
Tip: If you plan to use the provisioning function, or if you have more than one Spectrum Control server in your environment, consider using dedicated user IDs and SSH keys per Spectrum Control server. This way, you can easily perform the following tasks:
•Delete the ID or change the password to prevent a Spectrum Control server from accessing a device.
•Identify the commands that are run by a Spectrum Control server by reviewing the audit log files of a device.
|
6. When you have entered the required information, click Next. Spectrum Control attempts to connect to the device.
7. After Spectrum Control successfully connects, you see configuration options similar to what is shown in Figure 4-3 on page 75.
Figure 4-3 Initial device configuration
8. When you have provided all the information and adjusted the scheduling based on your needs, click Configure.
Performance data collection begins after a successful initial probe.
|
Notes:
The 1-minute performance monitoring interval facilitates finer granularity for troubleshooting, debugging, and reporting.
Be sure to consider the performance and space usage when running 1-minute performance monitoring on your server. The space usage is DB2 space usage. The size of the data that is collected is related to the number of objects that the devices contain (for example, LUNs and disks).
Here are some limitations of 1-minute performance monitoring:
•The 1-minute intervals show only in a chart if the complete time range has 1-minute intervals; otherwise, it “falls back” to 5-minute intervals.
•In Cognos, the 1-minute intervals are not available.
•The TPC exposed views do not support 1-minute intervals.
The alerting is evaluated based on 1-minute intervals, so you might need to think about adjusting your alert thresholds (no 5-minute averages, so the threshold might need to be set higher, with longer intervals between repeating alerts).
|
For the full set of administrative actions for storage systems, see the Spectrum Control IBM Knowledge Center, found at:
4.1.2 Adding storage systems from other vendors
Non-IBM storage devices are typically connected to Spectrum Control by using a CIM agent. Complete the following steps:
1. From the Add Storage System window that is shown in Figure 4-1 on page 73, click the All others device icon.
2. Specify the connection information for your device type.
3. Complete the configuration steps in step 7 on page 74.
For the full set of administrative actions for storage systems, see the Spectrum Control IBM Knowledge Center, found at:
4.1.3 Adding fabrics and switches
SAN switch and fabric configuration options differ by switch vendor. The Add Device wizard guides you through choosing the correct options:
SNMP Agents SNMPv1 or SNMPv3 can be used to communicate with all vendor switches.
CIM Agents Spectrum Control can communicate with CIM Agents for all switches except Cisco. The CIM Agent is the preferred method for Brocade switches.
|
Tip: The CIM agent is the preferred method for Brocade switches because it collects the richest set of data. If both agent types are configured, Spectrum Control uses the CIM Agent to collect data.
The SNMP agent is the only method available for Cisco switches and provides full functionality.
Note: The Storage Resource agent does not collect fabric information.
|
The Spectrum Control IBM Knowledge Center explains what information is collected by each agent for each switch vendor. It can be found at the following website:
Adding switches to Spectrum Control
To add one or more switches to Spectrum Control, complete the following steps:
1. Check the support matrix for supported switches, firmware versions, and SMI Agents before you start adding devices, which is found at the following website:
Click Spectrum Control 5.2.x in the column labeled Switches.
|
Tip: The preceding URL will not change when new Spectrum Control versions are released, so it is a good idea to bookmark the link.
|
2. Add the switch by clicking Network → Switches.
3. Click Add Switch to add a switch and its data source.
4. A dialog box opens (see Figure 4-4) in which you select the type of device that you want to add. By selecting the type of device, the agent configuration options are customized.
Figure 4-4 Select the type of switch you are adding
5. Configure the connection to the switch based on the vendor and the available agent options:
a. For Brocade switches, you should monitor with Network Advisor as the CIM Agent. Change the default communication protocol, port, and namespace if needed for your environment. When you have entered the information, click Next to continue, as shown in Figure 4-5.
Figure 4-5 Adding Brocade switches with Network Advisor
You can add the Brocade switches by using SNMP agents, but it is not necessary if you are using Network Advisor as a CIM Agent.
|
Tip: You can download and install the Network Advisor Software for Brocade switches without requiring a license and incurring additional costs. During the Network Advisor installation, select the SMI-S Agent only option.
|
b. For Cisco switches, the only option is to add them by using SNMPv1 or SNMPv3. Select the protocol and then enter the connection information and the IP addresses of the switches to add. Click Next to continue, as shown in Figure 4-6.
Figure 4-6 Add switches by using SNMP
|
Tip: All of the switches that are added in this window are added by using the same connection details. If your switches have different user names, passwords, or community strings, you should add similar switches together and then click Add Switch again for the next set.
|
6. Spectrum Control verifies that it can communicate with the data source and then discovers which switches and fabrics are available through this data source.
7. Next, probe and performance monitor schedules are configured. When you have provided all the information and adjusted the scheduling based on your needs, click Configure, as shown in Figure 4-7 on page 79.
Figure 4-7 Configure probe and performance monitor schedules
Probes
Network resources for Brocade are almost always probed on the fabric level, but are probed at the switch level for Cisco. Brocade NPV (access gateway) switches are the exception. They are probed on a per-switch basis, and not as part of a fabric probe. For all other Brocade switches, if you run a probe for a single Brocade switch in a fabric, it is the same as running a probe for the Brocade fabric. The same is true when you schedule the probes or look at the log files. You can do these activities for both Brocade and Cisco from either the context of a switch or the corresponding fabric.
Although Brocade probes are typically at the fabric level, performance data collection is performed at the individual switch level.
4.1.4 Next steps
The Add Device wizard simplifies the process of adding devices to Spectrum Control by including the scheduling setup and customization. To complete the process of adding devices, here are several tasks that you should complete:
•Wait for the probe to finish to see whether there are any components with a status other than Normal that you should acknowledge (for example, ports that are not connected to a fabric).
•Review names, Custom Tags, and Location in the property window of the device, and update them as necessary.
•If you added a storage system, you might want to continue with setting the tier levels for the storage pools and add the storage pools to capacity pools.
•Review retention settings, job logs, performance data, and removed resources. Click Settings → History Retention in the menu bar to review or edit them.
•Check the default alerts and set alerts for events or thresholds and specify email addresses or other alert notification options. For more information about alerting in Spectrum Control, see Chapter 6, “Advanced alert management” on page 177 or the Spectrum Control IBM Knowledge Center, found at:
4.1.5 Updating and testing a data source
A data source is the link between Spectrum Control and a device. Over time, it is not uncommon to expect that the credentials (user ID and passwords) will need to be updated. In some cases, the IP address might need to be changed.
Testing the connection
It is important to know whether Spectrum Control can communicate with a device. To test the connection, complete the following steps:
1. From the list of devices (or from the devices detail window), start the test from the Actions menu by selecting Connections → Test Connection.
In Figure 4-8, we used the context menu because using the Actions menu would have hidden some more important parts of the window.
Figure 4-8 Test Connection window
2. Spectrum Control tries to use the data source to communicate with the device. You see a message indicating whether the connection test was successful. Click Close to return to the previous window.
If the test was successful, Spectrum Control updates the status. If you are working with a device that has multiple data sources that are configured, this menu and the Update Credentials menu have submenus that list the data sources, as shown in Figure 4-9 on page 81.
Figure 4-9 Update Credentials window
Updating the credentials
To update the device credentials, complete the following steps:
1. To update the user name and password, from the Actions menu click Connections → Update Credentials from either the device detail window or from a list of the devices.
2. Update the User name and Password fields as required.
The menu and the window where you enter the information looks slightly different from device to device because the interface depends on the data source type.
Updating the data source IP address
There is no explicit function to update the IP address for a data source. This section describes the way we simulated this scenario:
1. We swapped the cluster IP address and the service IP address of one of our SAN Volume Controllers.
2. We then added a storage system to Spectrum Control by using the new cluster IP address.
What happens in the background is that Spectrum Control runs a discovery and realizes that the device is known and simply updates the IP address. A message displays and indicates that the resource is being managed. When you see this message, click Cancel. The IP change is saved to the database. Spectrum Control now uses the new IP address.
4.2 Adding servers
You can add servers for monitoring by deploying a Storage Resource agent or by adding an agentless server to Spectrum Control. An agentless server is a server or virtual machine (VM) that is monitored by Spectrum Control but does not have a Storage Resource agent deployed to it. The agents are used to get a view of the connectivity between servers and other top-level resources, such as storage systems and switches.
4.2.1 Choosing an agentless server versus a Storage Resource agent
In some environments, deploying a Storage Resource agent to the server is not an option, either because of security restrictions in providing administration credentials, or simply to avoid loading a production server with agent code. An agentless server gives you the ability to model a host server, either a physical server or a VM, in Spectrum Control without deploying a Storage Resource agent.
By the definition of agentless servers, it is possible (as with Storage Resource agents) to get a view of the connectivity between servers and other top-level resources. This view is useful in troubleshooting scenarios to view performance data for connected subsystem and switch resources.
To gather server information, Spectrum Control correlates resources that are consumed by the server from storage system host connections and fabric zone aliases. The agentless server shows server Fibre Channel ports, HBAs, SAN-attached disks, and related SAN components. In many cases, you find that the agentless server provides the necessary information.
To determine which type of server to add, follow the guidelines in Table 4-1.
Table 4-1 Choose the type of server to add
|
Condition
|
Add an agentless server
|
Deploy a Storage Resource agent
|
|
View monitored resources to which a server is connected.
|
X
|
X
|
|
View storage assigned to a server from other monitored resources.
|
X
|
X
|
|
View the performance of the SAN-related resources connected to the server.
|
X
|
X
|
|
Organization restrictions prohibit deploying agents on servers.
|
X
|
|
|
View information about internal server resources, such as controllers and disks.
|
|
X
|
|
Note: Agentless servers do not provide file system or local disk information. Use Storage Resource agents to access that information.
|
4.2.2 Adding an agentless server
An agentless server can be any grouping of worldwide port names (WWPNs) on which you want to report, for example, a Cluster or Application.
Add an agentless server by providing the IP address or host name of the server. The following optional entries can be provided:
•OS type
•Location
•VM or physical server
•WWPN
|
Important: Although adding WWPNs is optional, without WWPNs, Spectrum Control cannot correlate the SAN resources that are consumed by the server.
|
The WWPNs are used by Spectrum Control to correlate the agentless server with the storage systems, fabrics, and switches that are being monitored. For example, the ports that you enter for an agentless server are compared to the ports on host connections for monitored storage systems. If there is a match between the server and a host connection, you can view the connectivity between the server and storage system and the capacity of volumes that are assigned to the server.
After an agentless server is defined to Spectrum Control, it is added to the Servers window (which can be viewed by clicking Servers → Servers), as shown in Figure 4-10.
Figure 4-10 Servers window
Use any of the following methods to add agentless servers for monitoring:
•Add a server by manually entering its information.
•Add a server from a host connection on a monitored storage system.
•Add a server from a discovered port that is connected to a monitored switch.
•Add a server from a discovered port that is connected to a monitored fabric.
The Add Server button is available in the Servers window, as are the panes for host connections and discovered ports. For more information about adding an agentless server by using these methods, see the Spectrum Control IBM Knowledge Center, found at:
4.2.3 Deploying a Storage Resource agent
You deploy Storage Resource agents if you want to enable full server monitoring for internal resources, such as controllers, disks, and file systems.
Open the Add Server window by clicking Servers → Servers → Add Server, as shown in Figure 4-11. Select the Deploy an agent for full server monitoring check box.
Figure 4-11 Deploy a Storage Resource agent
For more information about adding a Storage Resource agent, see the Spectrum Control IBM Knowledge Center, found at:
|
Note: The Storage Resource agent for Spectrum Control does not collect information about fabrics or switches. For the relevant agent options, see 4.1.3, “Adding fabrics and switches” on page 76.
|
4.3 Storage Resource agent use cases
This section describes administrative use cases that are based on monitoring the Storage Resource agents for Spectrum Control. They involve updating agent properties and handling changes to the DNS name or IP address for a server.
4.3.1 Updating agent properties
This section describes how to edit Storage Resource agent properties. Complete the following steps:
1. Open the Agent tab in the Properties window of the server, as shown in Figure 4-12 on page 85.
Figure 4-12 Agent properties
2. Click Edit to modify the agent properties. Several of the agent properties can be changed from the server Action list or the menu.
4.3.2 Changing the IP address or DNS name for Storage Resource agents
There are situations where you must make changes to the network setup that affect your Storage Resource agents. This section describes common network changes and how to update the Storage Resource agent to comply with the network change.
Changing the IP address of a Storage Resource agent
If you must change the IP address of a computer running the Storage Resource agent, you can do this change without any impact to Spectrum Control because Spectrum Control is always using the DNS name to connect to the Storage Resource agent.
Changing the DNS name of a Storage Resource agent
When you change the DNS name, Spectrum Control can no longer contact the agent. The new DNS name must be communicated to the Spectrum Control server. To update the Spectrum Control server, complete the following steps:
1. Stop the Storage Resource agent.
2. Create a file called REGISTERSRA (uppercase with no extension) in the root directory of the Storage Resource agent.
3. Restart the Storage Resource agent.
When the agent detects the REGISTERSRA file, it connects to the Spectrum Control server and registers itself, or simply updates the DNS name if Spectrum Control already knows that SRA.
This procedure is typically used when Spectrum Control agents are included in a master image that is used to install servers. For more information about the procedures, see the Spectrum Control IBM Knowledge Center, found at:
|
Note: You can use the same process to register a Storage Resource agent with a new Spectrum Control server when you modify the connection information in the configAgent.config file.
|
4.4 Additional configuration tasks
After you have the Spectrum Control software successfully installed with resources added and are successfully collecting data, there are still configuration tasks to help you understand and use Spectrum Control. The configuration can be changed later if you want.
4.4.1 Installing additional applications
Table 4-2 shows additional applications and tools that can be installed to complete the implementation of the Spectrum Control environment.
Table 4-2 Further configuration items
|
Item
|
Description
|
|
IBM Data Studio
|
As of DB2 10, the Control Center is no longer available. The replacement for it is IBM Data Studio. The software is available at no charge and can be downloaded from the following website:
You do not need this software to run Spectrum Control. However, if you used the DB2 Control Center to do some basic administration tasks or to develop SQL queries for custom reporting, you should consider downloading and installing the Data Studio.
Data Studio is useful in troubleshooting Spectrum Control application slowdowns resulting from slow-running SQL queries. This tool can also recommend actions to speed up the performance of Spectrum Control (for example, adding indexes and reorgs).
The following links provide more information about Data Studio:
|
|
Mozilla Firefox
|
You might want to install Firefox on the Spectrum Control server. Normally, you use the browser on your computer to access Spectrum Control, but there might be situations where you directly log on to the server, so having a working browser is helpful. For more information, see the “Supported Platforms Agents, Servers and Browsers” topic, found at:
|
|
Adobe Flash Player
|
The View Data Path function requires Flash Player be installed.
|
4.4.2 Backing up Spectrum Control
It is important to have a backup strategy for your Spectrum Control environment. There can be several items that you need to back up depending on how you use Spectrum Control. Here is a list of items that you should include in your backup strategy:
•Spectrum Control database
•Tivoli Common Reporting/Cognos Content Store
•Reports that were saved to the file system
•Tivoli Common Reporting/Cognos Report definitions
Because Spectrum Control and the Tivoli Common Reporting/Cognos Content Store are both stored in DB2, the backup can be easily integrated with Spectrum Control or other backup tools. Any reports that are saved in the file system should be part of a regular file system backup.
The backup of report definitions is more complex. Because any authorized user can create a report any time, you should set up a job in Cognos to create an export of the report definitions regularly. The export is a compressed file that is written to the local file systems so that it can be backed up by the regular file system backup.
For more information about the backup of Spectrum Control, see IBM Tivoli Storage Productivity Center Beyond the Basics, SG24-8236.
|
Important: For this approach to work, you should not use the “My Folders” location for your reports. In our testing, we did not find a way to export regularly reports that are stored there. As a preferred practice, create a folder in the Public Folders tab, and save your reports there.
You should not save any of your reports in the folders that Spectrum Control is creating. The reason is that this might make future upgrades more complex than necessary. If you have a folder that is created in the Public Folders tab, you can create links to the Spectrum Control provided reports, which makes it easier to find and go to the reports.
|
4.4.3 User authentication
Spectrum Control supports a series of federated repositories: Local OS, LDAP, or AD. LDAP and AD are external authentication methods, and Local OS is not.
The web-based GUI allows most actions to occur without requiring you to log in to the server locally. So, the use of external authentication servers is growing.
Depending on your specific environment, there are different methods for authentication in a Spectrum Control environment, as shown in Figure 3-1 on page 48. By default, Spectrum Control uses Local OS authentication for the web-based GUI.
|
Note: Tivoli Common Reporting/Cognos Reporting uses a default file-based repository with the JazzSM package. The Cognos GUI checks the user ID and password that is entered against a file that contains user IDs and passwords instead of checking with the local OS. Therefore, you must use the same user ID and password that you entered during the JazzSM installation to access Cognos.
|
4.4.4 Tivoli Common Reporting/Cognos Reporting single sign-on
Even when Tivoli Common Reporting/Cognos Reporting is set up by using the same authentication server as Spectrum Control, you still must authenticate again when you start the reporting.
There is a way to configure the WebSphere Application Server server running Tivoli Common Reporting/Cognos Reporting to accept and use single sign-on (SSO) tokens so that you do not need to enter a user ID and password. For instructions about how to set up this form of authentication, see the following website:
|
Important: Single sign-on works only if you have switched from the default file repository authentication to LDAP in the WebSphere Application Server server that Cognos is using.
|
4.4.5 User management in Spectrum Control
Roles determine the functions that are available to users of Spectrum Control. When a user ID is authenticated to Spectrum Control through the GUI, CLI, or APIs, membership in an operating system or LDAP group determines the authorization level of the user.
There are three roles that are available in Spectrum Control, as listed in Table 4-3.
Table 4-3 User roles in Spectrum Control
|
Role
|
Users that are assigned this role
|
|
|
Administrator
|
Can use every function in Spectrum Control. By default, the following groups are assigned the Administrator role:
•Windows: Administrators
•Linux: root
•AIX: system
|
|
|
External Application
|
Cannot log in to the Spectrum Control GUI. This role should be used for external applications that use the Spectrum Control provisioning functions, such as:
•vSphere Web Client Extension for Spectrum Control
•IBM SmartCloud Storage Access
|
|
|
Monitor
|
Can log in to Spectrum Control but cannot run any function. A user with this role still can see all information and open log files, but here are the only actions that the user can do:
•Acknowledge alerts
•Acknowledge a non-normal status
•Set the tier level of a storage pool
|
|
You assign a role to a group of users, so all users in that group can perform certain actions in Spectrum Control. The role to group mappings are available by clicking Settings → User Management from the menu bar. Figure 4-13 on page 89 shows the User Management window where user groups can be configured.
Figure 4-13 Spectrum Control User Management
From this window, you can also directly start the WebSphere Integrated Solutions Console, which is where you configure the authentication mechanism, such as AD or LDAP.
For more information about user management, see the Spectrum Control IBM Knowledge Center, found at:
4.4.6 Preferred practices configuration tips
This section provides tips about how we typically configure a Spectrum Control environment and the benefits of doing it this way.
User IDs for device logins
It is easier to use a superuser ID when you add a device to Spectrum Control than create a user ID that is specific for the storage management task. However, as a preferred practice, create a special user ID for Spectrum Control on the storage device. When you have two Spectrum Control servers (one for test and one for production), consider using separate users IDs for both Spectrum Control servers so that you have more control over which Spectrum Control server can work with a specific device.
In our lab environment, we created a user who is called tpc_87 for the Spectrum Control server whose IP ends in 87 on the devices that we added to that Spectrum Control instance.
For some data sources, this task requires extra effort because some CIM agents do not require authentication. Because Spectrum Control is often the only application that uses the CIM agent, we did not go through the extra effort of adding that user for the connection between Spectrum Control and the CIM agent.
The IBM System Storage DS4000® and IBM System Storage DS5000™ CIM agent is an example of a data source where you do not configure a special user ID setup. Typically, use the user ID any and enter the password any when you configure the data source to document that no special user ID is set up between Spectrum Control and the CIM agent.
For the communication between the DS4000 or DS5000 CIM agent and the device, create and use a special user ID.
|
Tip: The added benefit for doing this setup is that you can look into audit files and understand easily which commands Spectrum Control has run. Some devices log only commands that change the configuration. For those devices, you probably do not see any entries in the audit log unless you use the optimization, provisioning, or transformation functions.
|
If you are using SAN Volume Controller or Storwize based storage systems and you do not want to or cannot use the user ID and password method, use a public/private key pair for each Spectrum Control server.
Naming conventions
Using a consistent naming scheme is a good idea. Here is one example of why this scheme can make your work so much easier.
The window that show information about pools does not provide information about the storage system type. If you have SAN Volume Controller or Storwize V7000 storage systems, you might want to see only those storage systems in the lists. The SAN Volume Controller and Storwize V7000 storage systems are like the pools that contain the volumes that are assigned to servers. So, the names of a storage system should include some information that you can use as filters in the GUI.
The filtering typically works across all columns in the displayed tables. You can also restrict Spectrum Control to look for the search string in a particular column, as shown in Figure 4-14, for example, for the one that is labeled Storage System.
Figure 4-14 Filter on Storage System
|
Tip: When the search function is used without restrictions, it also looks for the search string in columns that are not displayed.
By default, the Custom Tag columns are not displayed, but because Spectrum Control looks into those columns anyway, you can use the Custom Tags to customize your environment in a way that lets you use filters to focus on what is important to you.
|
Because the values in a name depend on your environment and requirements, there is no general schema to use. Also, you might have well-established names that you do not want to change. The next section provides alternative ideas.
Adding information to Spectrum Control
Spectrum Control offers ways that you can customize it. These ways might be more applicable than naming conventions in some cases. Here is a list of items that you might find useful:
•Custom Tags/User-defined properties
These properties are the most versatile custom attributes that you can use for many different purposes. They are available for the following items:
– Storage Systems
– Storage Pool
– Servers
– Hypervisors
– Switches
– Fabrics
•Location
The location property can be set for the following items:
– Storage Systems
– Servers
– Hypervisors
– Fabrics
– Switches
– Capacity Pools
For custom reporting, you can find the properties in the data model within the “Component Properties” folder of the corresponding component. An example is the label “Storage Virtualizer Pool Custom Tag 1”.
•Tiers for storage pools
Because there are many ways that users define what a tier is, Spectrum Control offers a simple and flexible way to define what you think a tier is. Basically, you can choose a number 1 - 10 and tag a storage pool with this information.
The Tier property is available for any storage pool no matter what type of storage system to which the pools belong. Because it is a simple tag, Spectrum Control does not check for any inconsistencies. For example, you can define all your pools containing 15 K rpm disks as Tier 5, and all 10k rpm disks as Tier 3 independent from the storage system type.
For custom reporting, you can find the properties in the data model within the “Component Properties” folder of the corresponding component as “Storage Virtualizer Pool Tier” or “Storage Pool Tier”.
•Capacity pools
Capacity pools are part of the cloud configuration concept in Spectrum Control. They provide a way of grouping of storage pools. Essentially, they are a special “type” of Storage Resource Group, which can contain only storage pools. Another difference is that one storage pool can be assigned only to one capacity pool.
|
Note: Not all of the preceding properties are available within Tivoli Common Reporting/Cognos Reporting for creating your custom reports through dragging. Nevertheless, some of the information is available through the TPCREPORT views.
|
4.5 Dashboard administration use cases
Here are some administrative use cases that are based on monitoring the Spectrum Control dashboard that can help you keep a clean and easy-to-monitor dashboard. They involve handling the alert status, performance monitor status, and the condition of your top-level resources.
4.5.1 Viewing the quick status on the dashboard
As a storage administrator that is responsible for SAN switches and storage, you might be contacted by an application team reporting that they saw a loss of paths, added latency, or possibly both. It is your responsibility to find the source of the problem and resolve it.
With Spectrum Control, you can monitor both SAN switches and storage within a common application, which can be a time saver. However, you must keep your monitoring environment in a state such that problems are apparent and not masked by unimportant alerts, which involves making sure Spectrum Control is actively monitoring the devices that you need and keeping your server in a state such that you can identify issues. For example, if everything is always reported as critical or warning, you learn to ignore the alerts.
Figure 4-15 shows a dashboard view with a storage system warning and various fabric and switch issues in the upper half. The lower left corner shows recent alerts and a quick view of how your performance monitors are running.
Figure 4-15 Spectrum Control dashboard
|
Tip: If you are not familiar with the displayed icons, select the ? in the upper right menu bar to open a help topic for the dashboard that explains each icon. You can open help anywhere in Spectrum Control in this manner.
|
If your switches and storage systems are normally without error or warning, you quickly can see where new issues are occurring by using the dashboard. Handles the alerts and any warning or error conditions on a regular basis for the monitored resources, which helps you keep your dashboard clean.
4.5.2 Handling alerts: Critical and warning
You can find details about the alerts that are shown in Figure 4-15 on page 92 by clicking Alerts in the lower left corner to open the Alerts window, as shown in Figure 4-16.
Figure 4-16 Alerts
In this case, the graph at the top of the Alerts window in Figure 4-16 shows that the errors are all related to the performance category. As you look at the most recent ones, you can see that they are related to two switch resources and an internal resource for each occurrence, such as a particular port. To see more details, select an individual alert row and select View Alert from the Actions menu. Figure 4-17 shows the details for the Port Congestion Index alert from row one. From this view, you can see that the threshold was crossed multiple times, how often, and to what level.
Figure 4-17 Port Congestion Index alert
What do you do in this case? There are a few options:
•Resolve the issue that is causing the alert.
Drill into the linked resource and internal resource to find additional information, such as whether a port is reporting an error status. You may also start an external element manager for the resource to investigate further. After investigating, you might want to do one of the other alert actions in this list.
•Acknowledge the alert.
Some alerts in Spectrum Control are triggered by conditions that commonly occur and can be ignored. In such cases, you acknowledge these alerts to indicate that they were reviewed and do not require immediate resolution.
•Remove the alert.
Remove alerts when you no longer need to view them. By default, alerts are automatically removed based on retention settings that are defined on the History Retention page in Spectrum Control.
•Set a different alert definition.
Determine whether the current alert condition is appropriate. If there is nothing of concern with how the resources are running, you might want to set a different threshold to avoid unnecessary alerts. The Spectrum Control default values are a starting point. The alert definition can be modified by drilling into the affected resource that is listed in the alert and then clicking Alerts in the left navigation and then clicking the Definitions tab, as shown in Figure 4-18.
Figure 4-18 Alerts definitions for a selected resource
For more information about alerting in Spectrum Control, see Chapter 6, “Advanced alert management” on page 177, or the Spectrum Control IBM Knowledge Center, found at:
4.5.3 Handling device condition: Warnings and errors
Use the Spectrum Control dashboard to get an at-a-glance snapshot of the health of your resources. After identifying top-level resources that need attention, use Spectrum Control to identify the internal resources that are encountering problems. The upper left corner of Figure 4-15 on page 92 shows a storage system warning condition, which is an example for how you can relate the condition value for a device to the status values of the internal resources that lead to this condition. The relationship between condition and status works similarly for storage systems, servers, hypervisors, fabrics, and switches.
Condition represents the most critical status that was detected on a resource itself and on its internal resources. For example, if an error status is detected on a disk, the overall condition of its storage system is error. Status represents the status of a resource as reported by its own hardware. For example, if a disk on a storage system is disconnected, an error status is reported for that disk by the storage system.
In this case, complete the following steps:
1. Drill into Storage Systems to view the listing of your storage systems and their conditions. Figure 4-19 shows the entry for the storage system with the warning condition.
Figure 4-19 Block Storage Systems listing
2. Double-click the storage system row to open the detail window for it. Figure 4-20 shows that the Pools and Managed Disks under the Internal Resources for this storage system are listed with a warning status.
Figure 4-20 Storage system detail window with internal resources with a warning status
3. The pools and managed disks are the resources contributing to this condition and should be reviewed. Drill into each of them to get more information. Figure 4-21 shows the pools with a warning status, and Figure 4-22 on page 97 shows managed disks with a warning status.
Figure 4-21 Pools with a warning status
Figure 4-22 Managed disks with a warning status
Both the managed disks and the pools listings indicate a warning status because of degraded ports. You can use this information to investigate further and resolve your issue. However, you might not want to resolve it now, but do not need the condition for that resource reflected in the dashboard view, does not continue to influence the top-level condition. This situation can make it easier to clear out warnings and errors that do not matter to you.
Figure 4-23 Acknowledge a condition for a top-level device
5. You also can select individual resources, such as managed disks. Right-click and select Acknowledge Status to be more selective about what you choose to acknowledge, as shown in Figure 4-24.
Figure 4-24 Acknowledge a status for an internal resource
After you acknowledge the storage system warning condition that is shown in Figure 4-24, the dashboard no longer reports a warning condition. The acknowledged condition does not contribute to the calculation like the warning condition did. Figure 4-25 shows the cleaned-up storage system section on the dashboard. The overall condition is now shown as normal.
Figure 4-25 Storage system condition acknowledged
|
Tip: If you acknowledge a condition, status, or alert, you have a similar option to unacknowledge it.
|
Resolving the device condition warnings and errors for switches and fabrics is a similar process where you open the listing for Fabrics or Switches. Find the resource with the warning or error condition and then drill into the internal resources for it to find their status values. As shown in Figure 4-26 on page 99, if you click Ports from the left navigation, you can see that the ones marked as warning are because of a degraded status. The affected switches and their port numbers are listed. It also indicates that the ports are disabled. This information can help you tackle this issue. Acknowledging the status values for these ports so that it does not contribute to the overall condition is an option too. If internal resources are listed with an error or warning condition, they contribute to the condition of the top-level resource. Work your way into each resource category to understand and resolve what you see.
Figure 4-26 Fabric ports with warning status
4.5.4 Clearing unreachable switches
Now that the Storage Systems section is cleaned up, as shown in Figure 4-25 on page 98, you can turn your attention to the switches and fabrics. Looking at the dashboard, there are two unreachable switches and fabrics, which are shown on the right. An unreachable fabric is typically caused by an unreachable switch or data source, such as a CIM agent or SNMP agent. The device might be unexpectedly unavailable or unavailable as the result of environmental changes or a plan to decommission hardware. In any of these cases, Spectrum Control should be cleaned up to remove the old data sources or correctly configure them.
Complete the following steps:
1. Start by drilling into the switches or fabrics listing. Figure 4-27 shows switches with several different conditions. The two switches with unreachable condition are used in this example. You might notice that the two unreachable switches have the same name, tpcf-b155. This switch is a Cisco physical switch (black switch icon) and its logical switch (blue logical switch icon). Both use the same data source connection to the SNMP agent and either one can be used for the next actions.
Figure 4-27 Switches listing that shows conditions
2. Right-click one of the unreachable entries and select Connections → Test Connection to check it. If it can be reached, running a probe is suggested, as shown in Figure 4-28. If you select to start the probe, the GUI is updated sooner.
Figure 4-28 Successful test connection
3. If the test connection does not pass, the data source might not be available. You can check and update the credentials. Right-click the switch and select Connection → Update Credentials. You can make any adjustments that are needed to reestablish a connection.
4. If you know that the switch is decommissioned or you no longer want to monitor it with Spectrum Control, you can right-click the switch and select Remove. This action removes the resource and all associated data, including probe and performance data.
Whatever method you use to clean up the unreachable switches, it is preferable to do so because unreachable switches are not being monitored.
4.5.5 Handling performance monitors
Performance monitors are an integral feature of Spectrum Control and its function. As an administrator, you should know whether your resources are being monitored so that the performance data that you need is available when you need it. The lower left corner of the dashboard, as shown in Figure 4-29 on page 101, provides a quick view to check the status for the defined performance monitors.
Figure 4-29 Dashboard view with Performance Monitors in the lower left
Ideally, the performance monitors are running without problems. If they are not, you should evaluate them further. In the case that is shown in Figure 4-29, most are running well, but there is one failed monitor and two that are not running. They are not actively collecting performance data for their resources.
Complete the following steps:
1. Select the Performance Monitors heading to see a listing of all performance monitors that are defined in Spectrum Control, as shown in Figure 4-30.
Figure 4-30 Performance Monitors status
2. The column that is labeled Latest Error in the performance monitors listing in Figure 4-30 on page 101 shows the error code HWNPM0291for the resource in the first row. It also shows that the start and stop times are close together for the one with a failed status. It did not run for long. To find out why a performance monitor failed, right-click its row and select Open Logs.
3. Scroll to the bottom of the log to find the reason that the monitor failed. Figure 4-31 shows that in this case that the monitor failed because the switch has no ports. It might be that a performance monitor should not have been defined for this switch, that the switch is incorrectly configured, or that something unexpected happened to the switch ports. Whatever the reason, you now have information that you can use to resolve this failure.
Figure 4-31 Performance monitor log failure
4. Figure 4-30 on page 101 also shows two performance monitors with a disabled status. You might want to disable these particular performance monitors, but it might also be that they were not scheduled when the resource was initially added to Spectrum Control. To enable a disabled performance monitor, right-click it and select Schedule. Set it as Enabled and select a data collection frequency, as shown in Figure 4-32. A similar action can be done to disable a performance monitor.
Figure 4-32 Enable a performance monitor
From time to time, you might see performance monitors report a status of “Running with problems”. If the performance monitor can recover, it goes back to Running. If it cannot, it reports a status of Failed. Figure 4-33 is an example of what you might see in the performance monitor log for a resource that is running with problems. One collection interval was successful, followed by a failure, and then followed by a successful one.
Figure 4-33 Log example of running with problems status
|
Note: You can view performance monitors from the individual resources in other views of Spectrum Control.
|
4.6 Fabric administration use cases
Here are administrative use cases based on monitoring switches and fabrics with Spectrum Control. They include monitoring inter-switch connections, the performance metrics for inter-switch connections, and N_Port Virtualization (NPV) mode switches connected to a fabric.
4.6.1 Viewing inter-switch connections
Inter-switch connections include the following types of connections:
•Inter-switch link (ISL)
•NPV Link (a single link between a native fabric switch and an NPV switch)
•F_Port trunk (also known as an F_port channel)
•ISL Trunk (also known as an E_port channel)
•Port channel
•Inter-chassis link (ICL) trunk
Inter-switch connections are viewable from both the switch and fabric details views. These connections may be between fabric switches, NPV mode switches, or access gateway mode switches. Two types of inter-switch connections, ISL and ISL Trunk, are displayed in Figure 4-34 as an example. You can drill into the links or the connected switch. The number under the Links column provides a reference to how many links make up that connection. An ISL is a single link, and an ISL trunk can contain multiple links between the switches.
Figure 4-34 Inter-switch connections listing
NPV link versus ISL link
An ISL is a single link between two native mode switches, and a single link between a native mode switch and an NPV mode switch is a specific type of inter-switch connection called NPV Link. An NPV Link may be a link between a Brocade native mode switch and a Brocade NPV mode switch or a link between a Cisco native mode switch and a Cisco NPV mode switch. Additionally, you may have a Cisco NPV mode switch connected to a Brocade fabric or a Brocade Access Gateway connected to a Cisco fabric. Both are supported options for NPV links. These single links are considered ISL or NPV Links only when they are not part of a Brocade trunk or Cisco port channel.
Figure 4-35 shows the properties for an NPV Link from the inter-switch connections listing for a native mode fabric switch. It provides details for the connected NPV switch and switch port. It was opened by selecting the number in the Links column for the NPV Link entry.
Figure 4-35 NPV Link shown from a native mode fabric switch
F-port trunk versus F-port channel
Whether Spectrum Control displays an inter-switch connection type of F-port trunk versus an F-port channel depends on the switch vendor, but the concept is the same for both. They are each a logical aggregation of links between a native mode fabric switch and an NPV switch. Brocade calls it an F-port trunk and refers to the NPV switch as an access gateway. Cisco calls it an F-port channel. Spectrum Control shows the specific type for the respective switch vendor.
When viewing a list of switches, the links are shown in a column on the right by default. The Properties window that is shown in Figure 4-36 on page 105 was generated by selecting the entry in the Links column for the switch named Testrisblast (the highlighted row in the background). Testrisblast is an NPV switch and the link that is shown is an F-port trunk.
Figure 4-36 F-port trunk properties from an NPV (access gateway) switch
For the F-port trunk in Figure 4-36, the NPV switch and the NPV switch port are shown along with the native switch and native switch port that make up the connection. Each row represents an individual link within the link aggregation for the trunk.
ISL trunk versus E-port channel and ICL trunk
Like the F-port trunks and F-port channels, Spectrum Control displays an inter-switch connection type of ISL trunk versus an E-port channel based on the switch vendor. Both of these types are a logical aggregation of ISLs between two native mode fabric switches, typically between two or more E-port links. Brocade calls the aggregation of these links an ISL trunk and Cisco calls it an E-port channel. Spectrum Control shows the specific type for the respective switch vendor.
Figure 4-37 shows the inter-switch connections for a Brocade native mode switch. From this view, the switches that are listed under the Switch 1 and Switch 2 columns indicate which switches are connected. You can select either of those switch names to drill into the switch itself for details, but what you can notice from this window is that the ISL trunk has two links while the ISLs each have one. The ISLs represent a single connection between two switches, and the Brocade ISL trunk or Cisco E-port channel is the aggregation of those ISLs.
Figure 4-37 Inter-Switch Connections showing ISL Trunk and ISL
For more information about this Brocade ISL trunk and its links, select the number 2 under the Links column in Figure 4-37 on page 105 to open the Properties window for the ISL trunk, as shown in Figure 4-38. From here, you can see all of the links making up the Brocade ISL trunk and the specific switch ports where they connect between each switch. From here, you can drill into any of the switch or switch port links for more information.
Figure 4-38 Properties for an ISL trunk
Additionally, there is another inter-switch connection type of ICL trunk that you might see in Spectrum Control. This is a logical aggregation of inter-chassis links (ICLs), which are the connections between core routing blades of two Brocade director switches. The core blades have special ICL E_ports that are used only for connections to core blades on other director switches.
Problematic links for inter-switch connections
Figure 4-39 shows an example of two different port channels that are configured on a switch that are recognized as being problematic. No connected switch is listed.
Figure 4-39 Inter-switch connections with problematic links
To understand why a connection is reported as problematic, click Problematic under the Links column for a particular connection, which opens a Properties window providing additional information about the port channel. Figure 4-40 shows that in this case the issue is because there are no ports in the port channel. You can drill into a list of associated ports to view more details or go to the management interface for the switch to configure the ports and resolve the problem.
Figure 4-40 Problematic port channel properties
Performance metrics for inter-switch connections
Performance metrics are available for inter-switch connections and can be displayed by clicking the Performance tab to display the information, as shown in Figure 4-41.
Figure 4-41 Performance tab for inter-switch connections
The performance tab for inter-switch connections works similarly to the performance tab for other resource types. The ISL or NPV Link data is the same as the port data because ISLs and NPV Links represent a port connection. A Brocade trunk or Cisco port channel is an aggregation of the data for the ports that are included in that connection. The performance metrics that are shown for a Brocade trunk or Cisco port channel are the same set of metrics as what is shown for ports.
As you view the information in the performance tab, you can go both up and down from the inter-switch connections. Figure 4-42 shows the options that are available as you right-click an inter-switch connection in the lower table. From here, you can choose to go to the port performance data for the entire switch or drill down into the performance data for individual switch ports.
Figure 4-42 Go to and drill down options for inter-switch connection performance data
If you choose the last option in the list from Figure 4-42, Connected Inter-Switch Connection Performance, you can go to the performance data from the perspective of the other switch participating in the inter-switch connection. If everything is working well on the two switches, the performance data for the inter-switch connections on each switch should be symmetrical.
For example, the send rates on one switch should match the receive rates on the other, and vice versa. However, you might see more errors being logged on one switch than the other, and that might help you identify a problem. Figure 4-43 shows an example view from the perspective of both switches that are involved in a Brocade trunk after selecting this option.
Figure 4-43 Viewing inter-switch connection performance data from both switches
|
Note: Performance data for a Brocade trunk or Cisco port channel is the aggregation of the performance data for the ports that are included in that trunk or port channel. If you choose to drill down to the switch port performance data for a simple ISL or NPV Link, the same information is shown because the ISL or NPV Link represents a single port to port connection.
|
Although performance data is not available for the fabric specifically, it is available for the switches within the fabric and you can view the data for the inter-switch connections or ports all at once.
Figure 4-44 shows the inter-switch connection performance data for a single fabric across multiple switches. The table below the chart lists the resources for which performance data is available in the selected time range. Two rows are displayed for each inter-switch connection, where each row represents the performance from the perspective of one of the two switches participating in the connection. (This situation assumes that both switches are part of the fabric and that both had performance monitors running. NPV mode switches are not considered part of the fabric.) The Switch Name column in each table row differentiates the switch perspective for the inter-switch connection that is listed in the Name column. The performance data is not identical for each side of the connection, but it is symmetrical. For a given trunk, the individual links that are members of the trunk should be load-balanced by the switches such that similar performance and throughput characteristics are seen.
Figure 4-44 Inter-switch connection information for a single fabric
|
Note: If the inter-switch connection is between a fabric switch and an NPV or access gateway mode switch, you do not see both sides of the inter-switch connection. The NPV mode switch is not part of the fabric and is not shown.
|
4.6.2 Viewing NPV switch ports and internal routes
In this case, an administrator wants to see where a server is connected to a fabric, particularly the connections for an NPV (N_Port Virtualization) switch. Figure 4-45 on page 111 shows the port listing for an NPV switch. Port 0 and port 1 are members of an F-port trunk, and information about that connection can be found in the Link column.
Figure 4-45 Ports listing for an NPV switch
The Internal Routes tab is available only for NPV switches and is shown in Figure 4-46. Click the Internal Routes tab to view the routes and connections from a node to a fabric switch as it goes through an NPV switch. Figure 4-46 shows the node ports that are connected to the NPV switch, which ports on the NPV switch to which they are connected, which ports to which traffic is routed, and the connections to the fabric switch ports. These internal routes show you the data path through the NPV switch.
Figure 4-46 Internal routes and data path from a server through an NPV switch to a native fabric switch
Gather additional information about any of the listed resources by drilling into the links.
|
Note: Performance metrics can be collected for F-port trunks and all other inter-switch connection types, as described in “Performance metrics for inter-switch connections” on page 108.
|
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.