Workload Analysis

Overview

Workload Analysis (WLA) is an integral part of SkySQL's Observability Platform:

  • Workload Analysis performs continual learning of prior and current events to form qualitative cohesive workload insights and narratives.

  • Learnings are automatically adjusted to the latest workload changes, identifying probable areas for proactive maintenance to maintain service quality unique to your workload.

  • Understanding the changes in typical workload patterns enables you to identify the types of atypical workloads that commonly trigger acute service problems like increasing resource constraints. Workload Analysis cuts through the complexity of understanding these variations and identifies them early while reducing the impact of human bias and making the analysis process repeatable.

Note

  • Workload Analysis is currently a Technical Preview.

Compatibility

SkySQL Services

  • Replicated Transactions

  • Single Node Transactions

Remote Observability Service

Workload Analysis is compatible with services running MariaDB Enterprise Server that use the Remote Observability Service.

Availability and Access

Workload Analysis is automatically available to Power Tier customers.

Database models are built every 24 hours for running services which have been running at least one hour.

To access Workload Analysis:

  1. Log in to MariaDB SkySQL.

  2. Access the "Workload Analysis" link in the main menu (left navigation).

Service List

The default view when accessing Workload Analysis is the Service List.

From the Service List, you can choose which of your SkySQL services to review.

Only compatible services will be shown.

After a service is launched, Workload Analysis requires a minimum of 2 weeks of data to learn and analyze your workloads. When a service is selected during this time period, a popup message is shown to the user.

Workload Overview

After selecting a service from the Service List, a Workload Overview interface will be shown.

The Workload Overview interface indicates the range of normal (green) to anomalous (red) activity that was identified within the workload on a specific date.

Tabs present specific details about the workload:

  • By default, the most recent week is shown.

  • Clicking twice on a date will show additional detail about the workload present on that date.

  • Clicking on a start date and an end date will show additional detail about the workload present in that date range.

Summary Tab

The Summary tab provides an overview of common database metrics for the selected time period.

A solid line indicates the observed values. Specific data points can be viewed by mouse-over on a chart. A dotted line indicates the expected value.

Metric

Description

TPS

The number of transactions processed per second.

QPS

The number of queries processed per second.

delta_Innodb_data_read

The amount of data read within a time interval. (InnoDB)

delta_Innodb_data_written

The amount of data written within a time interval. (InnoDB)

delta_Slow_queries

The number of queries which took longer than long_query_time (default 10 seconds) within a time interval.

Anomalies Tab

The Anomalies tab shows the most anomalous metrics for the selected time period.

Charts are shown for anomalous values of metrics, including Status Variables.

A solid line indicates the observed values. Mouse-over any data point to see the value of the metric and how that value deviates from the norm. A dotted line indicates the expected value.

Custom Tab

The Custom tab allows you to view the observed values of metrics, including Status Variables.

The "Type in metric name" box allows you to specify which metrics to display. Multiple custom metrics can be shown. Unwanted custom metrics can be removed using the 'X' icon.

Specific data points can be viewed by mouse-over on a chart.