MySQL Enterprise Monitor is available as part of the MySQL Enterprise subscription, learn more at http://www.mysql.com/products/.
This section highlights new functionality and is geared towards users of earlier versions of MySQL Enterprise Monitor.
If you are familiar with earlier versions of MySQL Enterprise Monitor, after one glance at the 3.0 UI you will immediately notice significant differences in functionality, organization and appearance. "Under the hood" changes are even more dramatic, with entirely different models and implementation for inventory, instruments, Query Analyzer, Advisor, Graphs and Event handling and notifications.
For information about upgrading MySQL Enterprise Monitor 2.3 to 3.0, see Section 7.2, “Guide for Upgrading to MySQL Enterprise Monitor 3.0”.
Policies, Groups and auto-scheduling are all improved to make administration of scale-out easier and automatic:
Add a MySQL Instance to one or more Groups using the MySQL Instances dashboard or when installing the Agent (which can be scripted)
Members of a Group automatically inherit Advisor and Graph schedules and configurations (all of which are auto-scheduled on startup), see Section 22.1, “Customizing Groups”.
Event Handling and notifications are centrally managed, group-aware and de-coupled from Advisor scheduling. See Chapter 18, Event Handling.
Adaptive scheduling - Changes in your IT infrastructure are auto-detected. For example, collections, analysis and notifications automatically adapt when new servers or slaves are provisioned, a master becomes a slave, file systems are mounted, etc.
As a result, connections to new Hosts and MySQL Instances can be configured without need for manual administration.
Zero Configuration Query Analyzer - Works "out of the box" with MySQL 5.6 Performance_Schema (supported by 5.6.14 or later).
Host monitoring - A newly-installed Agent begins collecting local CPU, memory, file system, and other OS-related data whether or not the Agent has also been configured to monitor a MySQL Instance at install time.
Auto-discovery of mysqld instances - Agent automatically detects local, unmonitored mysqld processes and reports them to the UI as unmonitored. You can ignore or add monitoring connections to them on the MySQL Instances Dashboard. You can fully automate monitoring of newly discovered Instances by editing the connection parameters for the MySQL Process Discovery Advisor and enabling "Attempt Connection".
Centralized Agent configuration - Add, update, ignore monitoring connections from the UI without having to tunnel into the host remotely.
Multi-instance monitoring - Conserve system resources on your monitored systems by installing a single Agent per host no matter how many MySQL Instances are running there.
Remote ("agent-less") monitoring - Installing an Agent on each Host is recommended, but only required for collecting OS-related data.
Trends, projections and forecasting - Graphs and Event handlers inform you in advance of impending file system capacity problems
Database Availability - SLA reporting is made easy with a graphical presentation of database availability for the past day, week and month
Highlights top problems - Ranks and presents Hosts and MySQL Instances with the most critical problems
Query Response Time index Numeric rating offers immediate insight into query performance (graphically presented by query, by Group or overall). For more information, see Section 23.2, “Query Response Time index (QRTi)”.
Visual SQL/graph correlation - Drill into any region on a graph to view SQL executing during the selected time period.
Expand any monitored Instance on the MySQL Instances Dashboard to browse its server configuration.
Retrieve and view a live map of Innodb Buffer Pool usage.
There are many additional UI improvements, including:
Dynamic page content updates using AJAX.
The Asset selector supports search (useful for large monitoring environments) and optionally show host and monitored assets.
Client-side graphing make graphs richer and more responsive, see Chapter 16, Reports and Graphs.
Most UI elements now include inline tooltips (hover to see their annotations)
Status Summary is displayed on
every page, updates dynamically, and shows current status
counters for Hosts monitored, MySQL Instances monitored,
MySQL Instances with invalid connection configurations,
Unmonitored MySQL Instances, and Emergency Events. The
counters are live links: click to navigate for details or
to resolve the issues they're reporting.
Advisor configuration and execution have been greatly improved, with changes including:
A new "emergency" level to highlight outages, etc.
False positives from flapping or spikes are avoided using exponential moving averages and other statistical techniques.
Advisors can analyze data across an entire group; for example, the Replication Advisor can scan an entire topology to find common configuration errors like duplicate server UUIDs or a slave whose version is less than its master's.
Advanced rules can search for multiple problem conditions but avoid cascading errors by generating only a single Event and notification.
The UI for Advisor scheduling and configuration centers around Groups and Group membership. Selecting a Group or individual MySQL Instance lets you view or customize its schedules, thresholds, etc. Any changes apply to both current and future Group members.
 Requires Query Analyzer, which now collects SQL performance data using the PERFORMANCE_SCHEMA from 5.6.14 or later. This release also supports Query Analyzer sources like the Connector plugins (C/Java, C/php, C/Net) or the MySQL Proxy. There are advantages and disadvantages to each approach, see Section 13.2, “The Query Analyzer” for a brief discussion of differences.
 Known limitation: Process discovery is not supported on Microsoft Windows.
 To edit, click Configure, select Advisors, open the Monitoring and Support Services Advisor category and edit. Known limitation: This action is limited to a single set of credentials.