[cosmos-dev] COSMOS i8 data collection/data visualization integration di

[Sheldon Lee-Loy <sleeloy@xxxxxxxxxx>]

Here is the meeting minutes from today
meeting:

Attendees:
----------------
Mark
Paul
John
Hubert
Bill
Leonard
Ali
Shivvy
Sheldon



At a high level there are two integration
points between the Data Visualization and Data Collection Component:

1. Outputters - java objects that contains binding and query logic to the data collection apis and produces JSON structures
2. BIRT ODA - object data access component that contain binding and query logic to the data collection apis that produces data sets used by report templates

I7 Integration Notes
------------------------------
In i7 the outputters utilize the following
data collection client interfaces:
org.eclipse.cosmos.dc.management.domain.client.ManagementDomainClient
org.eclipse.cosmos.dc.dataManager.client.DataManagerClient
org.eclipse.cosmos.dc.mdr.client.MdrClient
org.eclipse.cosmos.dc.broker.client.IBrokerClient

In i7 the BIRT ODA utilize a helper
methods defined under org.eclipse.cosmos.dc.birt.helper.LocalHelper.
 These methods contains the binding and query logic for the old query
capability data collection runtime.

I8 Integration Notes
------------------------------
Domain Manager/ Broker/Data Manager
Integration
===============================================
Design document for 208595 (http://wiki.eclipse.org/COSMOS_Design_208595)
discusses the outputters required to query the domain manager and broker
to populate the navigation tree.  
   -  It was decided
that we should only show a list of data managers to simplify the ui rather
than showing the management domain and list of brokers since the user will
most likely only be interested in the data managers
In i8 the data collection component
will maintain the i7 data collection client interface apis as defined in
the following classes:
org.eclipse.cosmos.dc.management.domain.client.ManagementDomainClient
org.eclipse.cosmos.dc.dataManager.client.DataManagerClient
org.eclipse.cosmos.dc.mdr.client.MdrClient
org.eclipse.cosmos.dc.broker.client.IBrokerClient

As a result little or no refactoring
is necessary to query the management domain, broker or data mangers.  
The IBrokerClient interface will require
an api to provide a list of Data Manager properties that the broker is
aware of.  This will allow the UI to determine if the Data Manager
is a logging, statistical or cmdbf data manager.  Specifically, the
broker will provide an api to get a list of Data Managers and associated
Data Manager properties.  One of these properties will be the capability
of the Data Manager to determine if the Data Manager has statistical, cbe
or cmdbf query capabilities.

Side note1: we need to take care of
cases where the datamanager is not a WSDM endpoint.  In these cases
the data manager may not have the ability to query for properties via WSDM.
 For i8 we are not going to tackle this issue.
Side note2: consider defining a generic
capability for String in and String out support

Logging and Statistical Data Manager
Integration
=========================================
In I8 the apis defined under org.eclipse.cosmos.dc.birt.helper.LocalHelper
will be separated into two provisional interfaces for statistical and logging
data.
These interfaces will be convinient
apis that a data manager will implement.  The following are methods
that the logging interface will implement:
        public
Collection<CommonBaseEvent> getAllCBE(String dataset) throws
Exception;
        public
Collection<CBESeverityCount> getCBECountBySeverity(String dataset)
throws
Exception;
        public
Collection<String> getDataSets() throws
Exception;

The following are methods that the statistical
interface will implement:
        public
Collection<StatisticalTuple> getStatDataForSource(String dataset,
String keyset, int
start, int
size) throws
Exception;
        public
Collection<String> getDataSets() throws
Exception;

        
Will refractor current report ODAs to
talk to the data mangers that implement the above interface.

For Logging and Statistical Data we
will utilize the existing database schema we have for i7.  We will
create sql statements to query the database and implement the above methods.
 An enhancement will be created to wrap TPTP common services around
a data manager.  This enhancement will be targeted to a future iteration.
 This will provide a better integration with TPTP large log support
and services.


Thanks,

Sheldon
______________________________________
Sheldon Lee-Loy
Tivoli Autonomic Computing, IBM Toronto Lab
email: sleeloy@xxxxxxxxxx
phone: 905.413.2610