pyslo 0.0.5

No project description provided

pyslo

Calculate service level objective measurements from metrics stored in common backends in accordance with the logic set out in the SRE Workbook

Getting Started

Installation process

pip install pyslo

Build and Test

pytest

Current Support

Providers

At this time, the Stackdriver backend is supported. Future plans include Prometheus and Azure Monitoring

Metric Types

Stackdriver

• Boolean

Logic

The library pulls raw timeseries data from the metric provider and performs aggregations in memory. This is in order to standardize the computation across providers.

Boolean Metrics

sli = good_events/valid_events

where

• good events = (sum of metric entries == True)
• valic_events = (sum of metric entries)

Examples

Google Cloud StackDriver

This example uses a metric provided by GCP for the Composer service.

from google.cloud import monitoring_v3
from pyslo.metric_client import StackdriverMetricClient
from pyslo.sli import Sli

PROJECT = <>

metric_client = StackdriverMetricClient(project=PROJECT)
metric_client.metric_type = 'composer.googleapis.com/environment/healthy'
metric_client.value_type = monitoring_v3.enums.MetricDescriptor.ValueType.BOOL

sli = Sli(metric_client)

sli.window_length = 30  # days
sli.slo = 0.99

sli.get_metric_data()

# Calculate sli/slo doing no group bys.
sli.calculate()
sli.error_budget()
print(sli.slo_data)

# Group by some metric labels
sli.group_by_labels = ['environment_name', 'project_id']
sli.calculate()
sli.error_budget()
print(sli.slo_data)