Transcript Document

Hoosh Afsar
Financial Engineer
Fannie Mae
Venkat Tummalapalli
Senior Product Manager
Oracle Corporation
Extending the reach of
Enterprise Manager 10g to
monitor custom applications
across the grid
Stove-pipe architecture is rigid and
does not scale well
Corporate Network
LDAP
over
SSL
Intranet
Firewall
Public DMZ
Internal
LDAP
Servers
Internal Logging Server
Extranet
Firewall
Intranet
Oracle9iAS
Others
Interfaces
Intranet
Clients
HTTP
over
SSL
Intranet
Listeners
Net*8
over
ASO
Intranet
Standby
Database
Internal
Applications
Secure External
Extranet Firewall
Listeners
Extranet
Oracle9iAS
9iAS
Intra
9iAS
Extra
Net*8
over
ASO
Logging
Server
(centralized)
Tuxe
do
Serv
er
Legacy
systems
Authentication
Server
Net*8
over
ASO
Intrusion
Detection
Extranet
Standby
Database
Net*8
over
ASO
HTTP
over
SSL
I
n
t
e
r
n
e
t
Our Computing Grid is flexible
Business Processes
Trade Valuation
Risk
Management
Financial Modeling
Portfolio Management
Analyze
demand for
resources
Simulation Engine Servers
Farm of 100+ machines
Adjust
supply of,
access to
resources
Our Grid needs an extensibile system
mgmt tool
Root-cause
Impact Analysis
SE Client
SEES
Client host
Enterprise
Server
SE-Agent
SE-Agent
Agent
Server
Trending
...
SE-Server
Computation
Server
Real time
alerting
Agent
Server
SE-Server
...
Computation
Server
Oracle Eco-System
Provisioning
Fannie Mae’s use of EM:
Assuring availability of a $1 billion/day
trade valuation system
SE
Applications
Reduce
Automate
Sun Fire 280R 8G RAM
Simplify
Scale
Repository
Oracle Enterprise Manager 10g:
Extensible solution for the Enterprise
1
End Users
Exhaustive individual
component
management:
Admin
Performance
Software Deployment
Oracle AS 10g
J2EE
2
Integrated Suite
Management:
Admin
Performance
Software Deployment
3
Integrated
Management of
Oracle Products
4
Critical
Management for
Oracle Eco-System
Oracle eBus Suite
Oracle Collab Suite
Web Services
Portal
Integration
Oracle9iAS
Top Link
OC4J Web Cache
Wireless
SSO
Oracle 10g
Custom
Applications
What Makes EM extensible ?
Zero barrier
to entry
Web Client
HTML
Management Servlets
Console API
JDBC
Open Management Repository
Repository
API
XML
Managed Targets
Standard API
What makes EM Extensible?
Rapid
Instrumentation
- Declarative definition of new targets :XML
You do not need to write oraTcl scripts or C data gatherer
catridges. Coding time is reduced by orders of magnitude : from
Months to Minutes
- Public interface to management data: Well documented
Repository schema
You will be able to use your favorite tools to slice and dice the
management data or to do ad-hoc reporting
- Support for Open Management Standards WBEM, JMX
What are the integration Points ?
EM Black Box
3
3. Extend Console UI
2. Build custom UI’s and
Report through Repository
Access
2
Agent
1.
Define and instrument
target metrics
1
Active
Management
Repository
Agent
Agent
Integration Investment Value Curve
Management Value for Customer
3
2
3. UI Integration
Higher cost – Most value
2. Repository Integration
Medium cost – Greater Value
1
1. Metric Instrumentation
Low cost -- Immediate Value
Integration Investment (Time and Resources)
Integration Points: Agent
 Function
–
Monitor the functioning of hardware or software targets.
 Public access via
–
–
XML
 Declarative discovery of target (targets.xml)
 Declarative definition of metrics (metadata for target)
Fetchlets
 data access mechanism
 SQL, Log file, OS, SNMP, URL Timing, …
 Instant value to your integration: Automatically shows
up in the EM UI without having to do any UI
changes!
Integration Points: Agent
Collection
Definitions
Collector
Task Executor
Transfer
Target
Metadata
Metric Engine
Fetchlet 1
Config
File
Fetchlet 2
Fetchlet 3
Java API
XMI
V 1.2
Managed Target
EM Agent
Collector
Metric Engine
Presentation
API
Task Executor
Etc...
Managed Targets
Preconfigured
WBEM Enabled
User Defined
V 1.1
Managed Target : Examples ..
Middle Tier
Apache 1
App 1
BIG IP
Backend
JS
JS
DB 1
JS
Cisco Router
Apache 2
JS
DB 2
Apache 3
JS
LOGIN
Server
JS
V 1.1
Integration Points : Repository
 Functions
–
–
–
–
Uploads data into the database from loader files
Rollups of raw data
Notifications for event occurrences
Global Inventory for tracking software patch levels
 Public access via
–
–
Views (ad-hoc reporting)
pl/sql api (procedural access, example add an
admin)
Management Repository
Central Policy
Application Level
Inventory
DB Config
Public “MGMT_” Level
Monitoring
Internal Table Level
ECM & APM
V 1.1
Integration Points: User
Interface
 Create homepages for your custom targets
 Plug in administration components that are
based off data collected in the repository.
 The administration components can directly
connect to targets or go thru the Agent to
make changes on managed node.
Putting it all together
Managed Targets
June 2000
EM Daemon
Internet
Real-time
Monitoring
HTTP(S)
XML
Target Definition
Files
Reporting
Management
Repository
Notifications
XML
Firewall
Metric Definitions
Metric Data
Alerts
Client Response Times
Web Server Response Times
Information
Rollup
V 1.3
Notification
Example: Simulation Engine Queues
SE Client
SEES
Client host
Enterprise
Server
SE-Agent
SE-Agent
Agent
Server
...
SE-Server
SE-Server
Computation
Server
Computation
Server
...
Agent
Server
How do I instrument my app ?
Prerequisite
• Separate the application into multiple targets, and
define them separately
• Determine what makes up the application:
 Components
 Availability
 Performance
How do I instrument my app ?
Workflow
• Define targets
- pick your metric
- compute metric
• Define collections
- collection schedule
- Instantiate your targets
Define target type
Define collections
Set thresholds
using the EM UI.
Declare the targets
How do I instrument my app ?
Agent
EM SDK Provides
Predefined Collection
Mechanisms: Fetchlets
Targets.xml
Collections.xml
<targetname>.xml
Define the targets you
want to monitor and the
metrics you want to
collect using XML
Example Fetchlets: SQL and
PL/SQL, SNMP, HTTP, Text,
WEBM…
Monitoring data
immediately available
from EM Console
Example : What makes up my app ?
Are my applications available?
Example: App specific metrics
Example: App specific metrics
Step 1: How do I define a target?
Step 2: How do I define collections ?
Step 3: How do I set thresholds ?
Step4: Declaring target instances
Step 5: Quality Assurance
 Test metrics using the EM Metrics Browser
 Setup collections into the repository
- Edit emd.properties and specify Repository
upload URL, Upload Interval
 The custom application is now ready to be
monitored by EM !
What did it buy for us ?
 The data that we collect will now be uploaded/ rolled
up periodically - Raw data -> hourly -> daily -> ..
 Time Series charts with various statistical data for
trending- Average, High/Low Watermark, Std Dev.
 Event detection and real time notifications
 Correlation across various layers of the FM stack
 Service level reporting
When will it be available ?
 1st planned release : End of CY 2003
Q U E S T I O N S
A N S W E R S