立即与支持人员聊天

获得即时帮助
完成注册

登录

请求定价

联系销售人员

Foglight for Virtualization Enterprise Edition 8.9.3 - Performance Tuning Field Guide

Characteristics of Poor Performance in Foglight Parameters that Affect Foglight Performance Critical Areas that Can Affect Foglight Performance Identifying Performance Problems

Hardware and Operating System Tuning

Running on Virtual Hardware

Performance Challenges CPU

Ready Time

Diagnostic Snapshot

Management Server Tuning

Dashboard Default Timeout

Java Virtual Machine Tuning

Getting Started Foglight JVM Configuration

-XX:+UseConcMarkSweepGC -Xmx and -Xms -XX:NewSize, -XX:MaxNewSize, and -XX:NewRatio -XX:MaxPermSize -Xss

Common Symptoms and Tuning Resolutions

Backend Database Tuning

Initial Database Configuration Settings

Monitoring and Managing Database Size Backup and Recovery Recommendations init.ora Configuration

Sample init.ora:

Database Maintenance Recommendations

Case

How the Management Server Uses the Database Index Management Memory Block Size Oracle Striping Oracle Tablespaces

Undo tablespace Custom Tablespaces

Microsoft SQL Server Tuning PostgreSQL Tuning Troubleshooting

Management Server Load Preventing Database Connections

High Availability (HA) Tuning

Tuning Connection Issues in an HA Implementation Managing Hosts with Multiple Network Interfaces JDK with IPv6 on Linux Management Server Automatically Restarted Other JGroup Related Issues and Information

Topology Changes and Topology Churn Canonical Data Transformations (CDTs) Agent Weight / Environment Complexity Large Topologies Sampling Frequency XML-HTTP Agent Adapter Dropped Agent Manager Log Messages Java EE Technologies

Appendix: Performance Health Check

Is the Server Getting the Right Amount of Memory?

Management Server Memory is Healthy if ... Management Server Memory Checks Possible Actions

Is the Server Getting Enough CPU?

Management Server CPU is Healthy if ... Management Server CPU Checks Extended Browser Interface Checks Possible Actions

Is the Database too Slow?

The Database is Healthy if ... Database Checks Possible Actions

Is the Database Growth Reasonable?

Database Growth is Healthy if ... Database Checks Possible Actions

Is There Too Much Data to Process?

Data Volume is Healthy if ... Data Volume Checks Possible Actions

Is the Model Stable?

The Model is Healthy if ... Model Checks Possible Actions

Are Too Many Alarms Firing?

The Number of Alarms is in a Healthy State if... Alarms Checks Possible Actions

Is the Business Logic Properly Tuned?

Business Logic is Healthy if... Business Logic Checks Possible Actions

Are There Too Many User Requests?

User Activity Is Healthy If... Possible Actions

Appendix: Analyzing a Support Bundle

Topology Sync Topology Limits

Diagnostic Snapshot

Memory Consumption Threads, Deadlocks, and Overall CPU Usage Useful Information ---- jboss.system:type=ServerInfo ---- jboss.jca:service=ManagedConnectionPool,name=jdbc/nitrogen ---- jboss.web:type=RequestProcessor,* ---- com.quest.nitro:service=Derivation

Derivation Related Issues

---- com.quest.nitro:service=Topology

Topology Related Issues

---- com.quest.nitro:service=DataCacheEviction

Analyzing a Performance Report

Server Rule Information System-wide Topology Changes JVM Memory Usage Management Server Garbage Collectors JDBC Connection Pool Derivation Rulette

Alarms Checks

•

Look at the total count on the Alarms page.

•

On the Alarms page, in the alarm table, change the alarm filters to include cleared alarms and observe the new alarm totals.

Possible Actions

•

•

Is the Business Logic Properly Tuned?

It is possible to configure Foglight® so that it is doing too much on the models. This usually happens as a result of massive misguided customization. Foglight’s customizability makes this easy to do. A set of derived metrics bound to TopologyObject can add literally thousands of expensive expressions.

Business Logic is Healthy if...

•

CPU on the system is low when no browser interface operations are initiated.

•

The number of rulettes and derived metric expressions on the system is in proportion to the number of topology objects. If there are more than 100,000 derived metric expressions, then the system is doing too much.

相关文档

The document was helpful.

选择评级

I easily found the information I needed.

选择评级

© ALL RIGHTS RESERVED. 使用条款隐私 Cookie Preference Center