Database Monitoring

Accelerator makes it easy to monitor your database and subcluster usage.

Subcluster Monitoring

A primary subcluster is identified by its default name, default_subcluster. In addition, the icon to the left of the subcluster name also identifies a primary subcluster. Subclusters without this icon are secondary subclusters.

Along with the default_subcluster icon, all subclusters have another icon to the left of their name representing whether or not the subcluster is running. If more than half the nodes in a subcluster are running, the icon is a blue checkmark (check_circle_outline). If not, it is an orange cross (cancel).

Node Status

The number of nodes, EC2 instance type, and start/stop status per subcluster can be viewed on the left-most (first) of the four subcluster status cards on the Home page. The card also has a visual representation of all its nodes’ statuses as filled circles. A blue circle (circle) represents a running node whereas an orange circle (circle) represents a stopped node.

Selecting the More link at the top-right corner of the card will open a modal with further information on the nodes’ IP addresses, status, and state.

  • IP address: the public IP address of the corresponding node.

  • Node status: the status of the instances as provided by AWS.

  • Node state: the latest state of Vertica on that node. The most seen states on Accelerator are:

    a. UP: When the Vertica database process is UP and running on a node.

    b. DOWN: When the AWS instance is running but the Vertica database process is not started yet.

    c. UNKNOWN: When the AWS instance is unreachable and Accelerator is unable to determine the state of the Vertica process.

Users Sessions

The second card of the four subcluster status cards on the home page shows the number of unique users and total running sessions per subcluster. It consists of:

  • Login: the earliest session the user is using.

  • Username: the user’s database username.

  • Open Sessions: the total number of database sessions currently open for the user.

  • Active Sessions: the total number of sessions (from the user’s open sessions) that are currently active, i.e., executing a job or query.

  • Client: the client application or programming language to connect with the database.

  • Connection: whether the Vertica connection is secure (SSL/TLS) or not.

Total Queries

The third card of the four subcluster status cards on the Home page provides a graphical overview of the total number of queries executed in that subcluster in the last 8 hours. Selecting the More link at the top-right corner of the card will open the Monitor page for that subcluster.

Note: The Total Queries graph represents the last cached data, i.e., the last time you accessed the Monitor page. For the latest data on performance of all live databases and subclusters, Vertica recommends using the Monitor page.

  • Hovering over the graph points will display the total number of queries at that time.

  • You can also click and drag on the graphical view to zoom into a specific area. When zoomed-in, a reset button will appear on the top-right corner to reset the graph.

Database Autoscaling Policy

The final (right-most) card of the four subcluster status cards on the Home page represents the Autoscaling policy for the database.

Autoscaling

  • “Elastic autoscaling” label followed by a blue “ON” chip indicates that the database autoscaling policy is set to elastic autoscaling.

  • “Schedule-based autoscaling” label followed by a blue “ON” chip indicates that the database autoscaling policy is set to schedule-based autoscaling.

  • “Autoscaling” label followed by a gray “OFF” chip indicates that the database has no set autoscaling policy.

Idle shutdown

  • “Idle shutdown” label followed by a blue “ON” chip indicates that database idle shutdown is enabled.

  • “Idle shutdown” label followed by a gray “OFF” chip indicates that database idle shutdown is disabled.

Log monitoring

Accelerator tracks and records all database and subcluster events for monitoring and troubleshooting purposes.

Event Log

The Event log presents logs in a tabular view with filters and sorting functionalities, letting you identify the required events with ease.

To view the Event Log, navigate to the Home page and locate your database. From there, expand the More menu in the top right corner of the database whose event log you would like to view, and select Event Log. From there, use filters in the table header to see the events.

Database event logs consist of:

  • Date & time of the event.

  • Event description. This includes:

    a. Create or terminate database

    b. Revive database

    c. Backup or restore database

    d. Create or drop subcluster

    e. Start or stop subcluster

    f. Increase or decrease subcluster size

    g. Edit instance type of subcluster

    h. Update CIDR range for database access

  • Database/Subcluster for each event.

  • Status of the event, i.e., Success or Failed.

  • Initiated by the User, Idle shutdown, or Elastic Autoscaling.

* Note: Since event logs are recorded and stored in the same S3 location as the database, deleting a database will also delete the logs. Event logs once deleted cannot be recovered.

Amazon Cloudwatch logs

If you configured Accelerator to receive events from Amazon Cloudwatch, Accelerator installs the CloudWatch agent in each instance and creates a CloudWatch log group in your account.

You can view Vertica log data for each database in your Amazon Web Services (AWS) account:

  1. Log in to your AWS account.
  2. Open the CloudWatch console.
  3. In the navigation pane, select Log groups.
  4. Search for your Vertica database log group. The log group name is the same as your database name.
  5. Under Log streams, filter with the Vertica instance ID. The log streams contain the Vertica logs for that instance.

For details about monitoring Vertica logs, see Monitoring log files. For details about Amazon CloudWatch, see the AWS documentation.

Usage Summary

The Usage Summary page gives you a summary of your vCPU usage and runtime to date. Here, you can monitor your usage on the database and subcluster levels, view high-traffic periods, and access session logs. This includes live, terminated, as well as deleted databases and subclusters over the specified times. This data can assist you in tracking the overall usage for your customer account group, and thereby validate your expense.

Access your database usage summary by navigating to the Usage page from the left-hand menu. From there, select the required database, find the required subcluster from ‘Subcluster Details’ and select View Session Log.The usage summary page is divided into two parts: Summary and Detailed Usage

Summary

The cards in the top portion of the usage summary page display an overview of usage in the last three months from the time of page access. The left-most card (first card) shows the overall usage for all databases in the last three months. All other database-specific cards show the overall usage for specific databases in the same time period.

Note: Timestamps on all database-specific cards indicate the time of first usage within the last three months.

Detailed Usage

Below the summary cards are usage breakdowns for all databases over a period of one month from the time of page access.

The graphical view provides an easy-to-read, subcluster-specific vCPU usage. Use the filters provided at the top-right corner to toggle between the different time periods.

  • Hovering over the bars will display the start and stop times.

  • You can also click and drag on the graphical view to zoom into a specific area. When zoomed-in, a reset button will appear on the top-right corner to reset the graph.

The tabular view below the graphs provides in-depth usage information, i.e., total runtime, total vCPU usage, and session logs, for all subclusters, along with subcluster configurations. Clicking the session logs link opens a modal with the following details:

  • Node name.

  • Instance ID.

  • Start and stop times for each node.

  • Runtime in minutes.

  • vCPU usage in minutes.

Current Contract Usage

View the Current Contract Usage section at the top of the Accelerator Usage Summary page to check how many vCPU hours you have used so far and how many hours you have left in your current contract period.

The panel on the left displays a bar graph showing the vCPU hours used since the start of the contract period, and either the number of hours you have left or the number of hours by which you have exceeded your contract. The middle panel lists the number of available vCPU hours and the number of days left in your current contract. The panel on the right shows the total number of vCPU hours provided in your contract, the number of vCPU hours you have used, and the date range covered by the current contract period.

Note: Only users with a paid subscription to Vertica Accelerator can view their Current Contract Usage.

Monitor

Vertica Accelerator facilitates monitoring the performance of all its live databases and subclusters. Monitoring can help in locating bottlenecks and identifying potential improvements to their configurations. It shows performance data from the last one month. The page is divided into two parts, as described in the following sections.

To view this information, navigate to Monitor in the left-hand menu of Accelerator, select the required database, and find the subcluster to be monitored.

Note: The Monitor page displays cached data. Use the manual Refresh button next to the page title to update the data. In case manual refresh fails to update the data due to conflict with other jobs, wait for a few minutes before refreshing it again.

Summary

The card in the top portion of the monitor page displays the overall query response time for all live databases. Use the filters provided at the top-right corner to toggle between the different time periods.

Detailed Performance

Below the summary card are usage breakdowns for all live subclusters over a period of one month from the time of page access. Use the filters provided at the top-right corner to toggle between the different time periods.

  • Hovering over the bars will display the start and stop times.

  • You can also click and drag on the graphical view to zoom into a specific area. When zoomed-in, a reset button will appear on the top-right corner to reset the graph.

For the selected time period:

  • Total Queries graphs depicts the number of queries executed on the subcluster.

  • CPU graph depicts the percentage of CPU usage per node.

  • Memory depicts the percentage of memory used per node.

  • Read from cache & communal storage can be used to analyze the difference between reads from the cache memory and the communal storage. Selecting the legend will toggle between the data being displayed.

  • Most active users provides a list of users with the most executed queries on that subcluster.