Skip to main content

Notice: This Wiki is now read only and edits are no longer possible. Please see: for the plan.

Jump to: navigation, search


COSMOS Main Page >

COSMOS Face-2-Face


Date: 04-December-06 Time: 1:30pm - 5:00pm PST Location: GroundWork Open Source, 139 Townsend Street, Suite 100, San Francisco, CA 94107

Call in 888 241 8547

pc 999451


Mark Weitzel

Craig Thomas

Sheldon Lee Loy


  • Given that we have Don (Data Collection) and Craig (Reporting UI) both possibly being available, then I suggest we begin to focus on the API b/t the two layers.
  • Establish the first target environment for the reports
  • TBD


These are notes from the meeting held Dec 4,2006, from 1:30-5pm PST.

Participating were Mark Weitzel, Sheldon Lee Loy, and Craig Thomas.

It is hoped that the participants will take a look at these notes, and make liberal edits to correct or re-cast them.

These notes are just the rough sketch of the conversation. The resulting updates to the Eclipse COSMOS end-to-end use cases and assumptions about the scope and availability of SML models, as well as the details of interface points are captured in the End-to-End use cases page.


  1. Scope of SML model. Options for simplifying assumptions. Concluded that a model of the J2EE application on Tomcat, behind an Apache HTTPD server, deployed on a Linux operating system, hosted on a commodity Intel 64-bit server with 4Gb memory and a Raid 5 disk subsystem is too complex for the March deliverables. Agreed that this model could be simplified to a J2EE application on an application server hosted on a computer. This simpler model has considerably smaller demands on SML, but offers the interesting dependency relationship, as well as the appeal to Eclipse developers of JMX instrumentation.
  2. The end-to-end use case seems to have 2 paths:
    1. Development time: Resource Modeling; Management Enablement; Data Collection; Reporting. All of these are conducted in one sitting. Further, the first two are the architecture and design tasks, while the second two are deploy, execute, observe tasks.
    2. Production time: Data Collection; Reporting. These two "downstream" activies are conducted on the live system. Rather than deploy, execute, observe, these can be thought of as "deploy, execute, monitor".
  3. End-to-end use case lifecycle and its relationship to SML-IF documents. Discussed that the Resource Modeling use case would result in the SML-IF documents for the app, the app server, and the host. In Management Enablement, we would add the triple-play: properties--how they are observed--how to take the measurement. This set of properties drives the generation of tooling in ME. In Data Collection.
  4. Moved quickly across the 4 stages of the end-to-end: resource modeling, management enablement, data collection, and data reporting. Narrated the responsibilities of SML and of the component code in each.
  5. Mocked up a report. The goal of the report is to assess the "state of a business that depends on applications". The report is organized as:
+ application-1 state: available
+ application-2 state: available
+ application-3 state: available
+ ...
the application node can be expanded, as can the application server node, as:
+ application-1 state: available
  + application-server-1 state: available
    - host-1 state: available
+ application-2 state: unavailable
  + application-server-2 state: available
    - host-1 state: available
+ application-3 state: available
  + application-server-2 state: available
    - host-1 state: available
+ ...

Back to the top