Thursday, April 17, 2008

Tivoli System Automation for Multiplatforms Overview

TSAM is not a new product, but it's definitely one that is becoming more popular these days, especially with the introduction of the Balanced Data Warehouse, which relies on its capabilities. So I thought a general overview might be useful. This high-level overview doesn't go into the gory details, as those are very well documented in the product documentation and in several redbooks and redpieces (links at the end of this post). Rather, this overview is meant to be a little more explanatory than the existing marketing material.

TSAM competes directly with Veritas Cluster Services (VCS). The purpose of both products is the same: to provide high availability to resources. A large portion of the functionality of TSAM is provided by Reliable Scalable Cluster Technology (RSCT) component, which is included as part of AIX, and is available for Windows, Red Hat and SuSE Linux. RSCT, along with its prerequisites (such as General Purpose File System), provides the mechanisms that allow you to define resources and have them monitored. There is no graphical interface for the configuration of RSCT or TSAM, so all of the configuration must be done from the command line.

RSCT provides the ability to "harvest" resources in your environment, so it will find and identify all of your network adapters, disks, filesystems, etc. You don't have to define *those* components. RSCT also provides the *ability* to define arbitrary logical resources and the *ability* to define automations to react to changes in the environment. RSCT also allows you to group resources into Resource Groups, which allow you to manage multiple resources as a single logical entity. RSCT also allows you to define Relationships between resources, allowing you to specify, for example, that these two filesystems must ALWAYS be available on the same node. (Update: these features are actually provided by TSAM) RSCT also allows you to define actions that should be taken when different resource change states. However, if you wanted to leverage RSCT directly, you would have to do a LOT of detailed customization for each component that you want actually automate to make highly available.

This is where TSAM fits in to make life easier for you. You've still got a good bit of customization to do, but it's a LOT less that you would otherwise have to do with the base RSCT. TSAM sits on top of RSCT and provides the mechanism that actually allows you to define arbitrary resources, resource groups and relationships, and based on your configuration (or "policy"), TSAM configures all of the automation necessary to make your resources highly available among the nodes in a cluster.

Specifically, TSAM allows you define resources of type "IBM.Application" (these are your applications) for which you configure Start, Stop, and Monitor commands. TSAM also provides the feature that automates the reflex actions required when your resources change state (for example, when an application goes Offline). With the introduction of these features (on top of RSCT as the base technology), TSAM fully automates the high availability of your applications based upon your configuration.

Much (certainly not all) of TSAM is scripts that are wrappers for the underlying RSCT commands. So when debugging TSAM, you'll primarily use RSCT commands. This is important to note, since it means that you must be intimately familiar with RSCT to be able to successfully deploy and manage TSAM. And TSAM makes a point of not replicating the RSCT documentation, so many of the commands you'll use to configure TSAM are documented in the RSCT manuals.

TSAM also provides an Operations Console, which is implemented as a WebSphere application that plugs into the WebSphere-based Integrated Solutions Console. This is really JUST an Operations Console, in that it ONLY allows operator-level management of your environment (e.g. bring an application online or offline, view information). This console provides you with NO configuration capabilities. You need to create all of your Resources, Resource Groups, Relationships, etc. from the command line. But it's not all painful. TSAM also provides the capability to export your "Automation Policy" (essentially, your entire configuration) to an XML file, and also to import from the same file. So this allows IBM and others to provide sample automation XML files that you can download, modify for your environment, and add to your existing policy.

All of the above is provided by the TSAM Base Automation Component. TSAM also provides an End to End (E2E) Automation Adapter, which allows you to manage and automate actions to be performed on resources across multiple clusters. This E2E component supports not only TSAM clusters, but also HACMP, Linux clusters, and even VCS clusters on Solaris.

TSAM also provides a TEC EIF publisher so you can send events to TEC (or the Omnibus EIF Probe) to allow you to monitor the status of your TSAM environment.


Lots of Documentation and Downloads:

Version 2.1 Redbook:

Main Product Page:

No comments: