As discussed in my previous post, this is a DRAFT of the Knowledge Share document I developed as a Helpful hints/information on APM for project managers. I am open for feedback and improvements.
CA Solution Knowledge Share
Janice R. Johnson notes
Product: Application Performance Management (APM)
- Basic Definitions
APM core components: APM Database, MOM, Collectors, Webview/Team Center, Agents
CA APM Component
Provides the ability to collect performance-related metrics from a JVM (via bytecode Instrumentation) or a .NET application.
Enterprise Manager MOM
Provides a mechanism to aggregate 1 or more Introscope Enterprise Manager Collector’s associated data store into a single logical view of the performance-related metrics.
Enterprise Manager Collector
Provides a mechanism to receive, retain, retrieve and monitor performance-related metrics from 1 or more Introscope Agents.
Cross-Enterprise APM Agent
Collects data from Non-.NET and non-Java applications.
WebView / TeamCenter
WebView is a browser-based interface to the dashboards, metrics, and other elements of the Introscope-monitored applications. It uses tabbed interface presenting information in these windows:
Console—shows data in dashboards, which contain Data Viewers.
Investigator—presents a tree view of Agents, applications, resources, and metrics.
Management Module Editor—presents a tree view of Management Modules and Elements.
TeamCenter provides an overview of an application environment. Administrators use Team Center to see the health state of the environment. Troubleshooters locate and triage problem areas in their environments.
· The dashboard shows the overall health of the environment.
· The map shows the relationships between individual components within the environment
APM integrates, collects metrics and alerts then sends them out to Spectrum or SOI (Services Operation Insight) or both…depends on your project and what APM is being integrated with.
- Basic Information for Kick Off/Client Meetings about the product
- APM must always go in a non-prod/test environment 1st…
- Testing out the agent (s) is important
- If integrating with SOI, SOI CA resource will create events and notifications in SOI for actions coming from APM. (It is important if you have a project that involves the integration of SOI or Spectrum or BOTH that the projects overlap to ensure the client understands the SOI/Spectrum requirements and integration points.)
- Important: The agents for APM must be deployed before any service modeling in SOI can begin
- Agents get deployed out and into the client’s environment
- Need client application owner or someone that has the right to write, change or restart server for the client application being integrated with APM
- Normally, we request read only rights for the CA Resource
- One agent deployment can be used for multiple apps on same server. For DotNet, this would be done automatically if it’s a standard ASP.Net app. For Java, each app would have to be instrumented and set up with or share an agent profile in their startup scripts.
- Agents are configured in a non-production environment and then you take that configuration to production
- Need client change management process to understand impact to timeline to go live and if an outage is required
- All ports for APM are configurable. Not a show stopper on which port to use
- APM has a couple ports it uses:
- 5001: is MOM <-> Collector communication
- 5443: is a Secure port to the MOM/Collectors, but it is not supported for MOM<->Collector or MOM<->WebView communication. The use case would be for Workstation/CLW<->MOM/Collector.
- 8080: WebView port, this is configurable to any non-reserved port, it can also be set for SSL connections, but customer will need to generate/provide security certificates (default 8443)
- 8081: Enterprise Manager Web/CEM port, this is configurable to any non-reserved port, it can also be set for SSL connections, but customer will need to generate/provide security certificates (default 8444)
- You may see a similar statement like this in your SOW…the above bullets give you background to explain or answer the clients question on “which ports apply to APM?
- From a SOW….
- Depending on the client security requirements, for example if using LDAP. Then you must make sure the MOM can get to the Active Directory
- There is a very tight integration between APM and SOI
- APM is deployed in non prod 1st and tested in non prod 1st then promoted to production because agents would add too much overhead in production
- Spectrum monitors/goes out and find servers/devices (H/W, routers at the H/W level)
- Spectrum asks where the hosts/agents are and then APM says here they are and then Spectrum is now monitoring all hosts
- Initial Resources Needed from the Client: Sys Admin, Database and Security
- The security team needing to do a security scan on the servers and to sign off on them.
- BASIC Diagram in Green
The MoM and the Collectors need to be on the same subnet for performance reasons.
--Thing Every Executive at the Client Site will want to know….When will I see the Dashboard?
2 phases of the Dashboard
- In Non-Prod: Need to get H/W in place, Collectors, APM up and running in Non-Prod and agents deployed (There is no data at this point but you can see the Dashboard)
- In Prod: You start seeing results on Dashboards. The thresholds are defined and alerts are active. This is when value is seen.
What is deployed first?
From a best practice we deploy the APM DB first because the MOM and Collectors connect to the DB as part of the install and while the processes are running.
The DB server not being ready isn't a complete blocker, as the MOM/Collectors/WebView can still be installed. Running the processes without the APM DB deployed though would result in error messages in the logs and is not a recommended configuration.