SLA overview
A Service-Level Agreement (SLA) is a key factor in determining whether a stress test is abnormal. During a stress test, monitoring SLA metrics for core services provides a clear view of your service's status.
Background information
PTS provides several industry-standard SLA templates for stress testing. For more information, see General-purpose SLA templates.
You can also customize SLA rules to meet your business needs and apply them to specific stress testing scenarios. For instructions, see Configure an SLA.
Terms
Before using this feature, understand the following PTS terms related to SLAs.
SLA
A Service-Level Agreement (SLA) is a key factor in determining whether a stress test is abnormal. During a stress test, monitoring SLA metrics for core services provides a clear view of the status of your service or architecture.
SLA metric
SLA metrics are the indicators used to monitor data during a stress test. They include business quality metrics and monitoring metrics for services such as ECS, RDS, and SLB. PTS will add more SLA metrics for other components, such as queues and SQL connections, in future iterations.
SLA rule
An SLA rule is an SLA metric combined with conditions that trigger an alert or stop a stress test.
SLA template
An SLA template is a collection of one or more SLA rules that are associated with specific industries.
SLA metrics
The following table describes the SLA metrics that are used to monitor data during a stress test.
SLA metric type | SLA metric item | SLA metric descriptions | Minimum monitoring granularity |
Business quality metrics | Response time (RT) | The response time of an API during a stress test. | 1 second |
Requests per second (RPS) | The number of stress testing requests sent per second. | ||
Success rate | The request success rate of an API during a stress test. | ||
ECS infrastructure monitoring metrics | CPU utilization | The percentage of CPU that is in use over a period of time. | 1 minute |
Memory utilization | The percentage of memory that is in use over a period of time. | ||
load5 | The average system load over the last 5 minutes. | ||
RDS infrastructure monitoring metrics | CPU utilization | The percentage of CPU that is in use over a period of time. | 5 minutes |
Connection utilization | Connections refers to the number of applications that can connect to an RDS instance. Connection utilization is the percentage of connections that are in use. | ||
SLB infrastructure monitoring metrics | Dropped connections | The number of connections dropped by a port. Statistics are collected from the IP:Port dimension. | 1 minute |
Abnormal backend servers | For example, the number of abnormal backend ECS instances. |