16.1 All Timeseries Graphs

Navigate to Reports & Graphs, and choose All Timeseries Graphs.

Note

While the main Overview page only shows connection related graphs for your MySQL instances, the All Timeseries Graphs page lists all graphs that relate to time.

The data shown in the graphs is determined by the server or group of servers selected in the server tree. Each graph is only displayed if there is appropriate data for the selected servers; for example, only slave servers have a Replication Slave Delay graph, and the Cluster graphs only apply to MySQL Cluster servers.

Graph Types

There are five main types of graphs:

  • Individual: A single Asset has multiple data sets graphed on a chart. For example, counts of SELECT, INSERT, UPDATE, and DELETE statements on a single Instance.

  • Combined: Multiple assets have a single data set, each graphed on one chart. For example, the count of selects for each of the five MySQL instances of a group.

  • Breakout: One (smaller) graph per Asset in a collection, showing one or more data sets on each individual graph. For example, one graph per CPU on a Host, or in a cluster.

  • Aggregate: One graph per collection of Assets, where the data sets across all Assets are combined via an aggregation operator. For example, one graph with each of the SUM(SELECT), SUM(INSERT), SUM(UPDATE), and SUM(DELETE) across the collection.

  • Treemap: A 2D hierarchical proportional-representation graph. See Section 16.2, “InnoDB Buffer Pool Usage” for an example.

Graph Behavior

Other behavioral notes about the graphs:

  • The graphs are context aware, and they display the most appropriate data for the Group or specific type of Asset (such as a MySQL Instance) that is selected. For example, when a Group is selected, details about the MySQL Instances in that Group will be presented in Aggregate. When a specific MySQL Instance is selected, its details will be presented directly. When a Host is selected, information about its File Systems will be presented in Aggregate, but when a specific File System is selected, it will be graphed individually.

  • The graphs show "item time spans" in large time ranges, and each data point represents a small time range. Because each graph is limited by the number of pixels per width, each data point must fit into the available pixels in the graph, and as such, the small time range adjusts to the graph size and range.

    For example, as a graph spans a larger segment of time, each data point also covers a larger segment of time. A data point could cover 3 minutes on a single day graph, or 6 minutes on a two-day graph, which affects the height and range. In this example, a data point could read as 3,000 on the first graph, and 6,000 on the second.

  • Hovering over a series item will isolate (focus on) the particular series. This might show a specific timestamp, or display a range of time where each data point represents multiple units.

  • Clicking a series item will remove it from the graph, and the X-axis and Y-axis values will adjust themselves accordingly.

  • Selecting a range (via mouse dragging) yields Query Analysis (QUAN) option for the particular range.

  • The Graph Height slider adjusts the height and size of every graph.

  • Graphs may be exported as CSV data, and may be viewed as either stacked or line graphs.

  • Group based graphs include a "Combined" graph type, which shows a series for every asset within that group (per host for host specific graphs, or per MySQL instance for MySQL specific graphs), rather than just the avg/min/max of all servers within the group. An example use case is for determining the cause of a spike.

  • Graphs can be reordered and filtered. For example, moving graphs to the top of a Graphs page allows them to be viewed first, and this reordering is preserved if the page is refreshed.

  • Under some circumstances, a graph may contain a gap, which could be caused by several different conditions:

    • After the Service Monitor was stopped, Agent specific metrics (such as the Database Activity Graph) can "expect" a 1 or 2 minute gap at the beginning of the offline period, and at the time the Service Monitor was started.

    • Some graphs, such as the CPU Utilization Graph, might have gaps between the time the Service Monitor is stopped and started if the Agent is running on the same host as the Service Monitor.

    • When it is impossible to evaluate data that is misaligned, so a gap is generated rather than misrepresenting the data.

All graphs support the ability to drag and select a specific area of the graph. When you select an area of the graph, the display adds a link to the Query Analyzer page to display the queries that were being executed during the selected time period. You can also export data for a particular graph by clicking Export as CSV while hovering over the graph you wish to export, or Export as PNG for a PNG image.