From time to time I get questions around Clustering with Operations Center. I decided to put together a Flash video that provides an overview.
To lead off the video, I'd like to provide some basic details. When users ask questions around Clustering, what they are specifically asking for is to set up Operations Center in an architecture that reduces the possibility of their end users experiencing an outage from our Console and/or Dashboard. The video covers the basics of the architecture, but there is more involved. Since Operations Center becomes the main (and only) Dashboard into the health and availability of the Enterprise, it is critical that this is highly available. After all, would you want your pilot flying blind without instruments? Operations Center is your instrumentation and proactive view into the health and availability of your mission critical services.
To be highly available, typically more hardware and software is required. The higher the uptime requirements, the more investment required. This is not unique to Operations Center, it is true to many solutions. Also, there are several solutions on the market that provide a "clustering" solution whereby a single device is split into multiple systems with a shared disk subsystem. Those solutions typically do not provide a Hot/Hot solution, for instance. When one of the systems crashes (was the HOT system), the other system (Cold or Warm as in, not actively running and up to date) must start up, which means, the users will experience some type of outage until the other system is fully up and running. These types of solutions will typically not provide a very high up-time percentage (e.g., %99.9).
Operations Center might not be the only thing that must be clustered, depending upon your requirements, other tools and/or technologies might require to be clustered. For example, if Operations Center is up and running, if it is a problem that when a point product like Netcool or Spectrum goes down and Operations Center no longer has visibility (e.g., the adapter disconnects and views go gray) into the current health indicators from the management tool, then potentially that tool requires clustering ALSO. Potentially the backend database that Operations Center uses to store information must be clustered also. Understanding your requirements will help in architecting a complete HA (Clustered) solution.
That said, watch the video; it provides an overview of the architecture of
Operations Center in a clustered configuration.
Aug 08 2012, 04:02 PM
Filed under: tips, dashboards, best practice, critical applications, Systems Management, IT management, Availability, BSM, Business Service Management, Live Dashboards, Operations Center, IT Operations, Systems Monitoring, BackUp, Business Services, Netcool, Load Balancer, High Availability, Tobin Isenberg, Spectrum, Data Center Solutions