Serverless Observability: where SLOs meet transforms

How do you build reliable SLOs when your serverless data is too massive to query? This talk shows how Elasticsearch transforms are the key.

#1about 3 minutes

Using transforms to solve scaling challenges in serverless

Elasticsearch transforms are used to create summarized indices from large datasets, addressing scaling issues during a serverless migration.

#2about 3 minutes

Understanding core SRE concepts like SLIs and SLOs

Service Level Indicators (SLIs) measure performance, while Service Level Objectives (SLOs) set targets to meet agreements (SLAs), with error budgets defining tolerance.

#3about 4 minutes

Exploring different SLO indicator types in Elasticsearch

A review of various SLO indicator types like APM availability, APM latency, and custom KQL helps in choosing the right metric for service monitoring.

#4about 3 minutes

How transforms power the underlying SLO architecture

The SLO architecture relies on a multi-layered transform service to roll up source data into summarized, entity-centric indices for efficient querying.

#5about 4 minutes

Creating transforms and setting up health alerts

Learn to create a transform through the Kibana UI and configure a transform health alert to monitor its status and trigger notifications via connectors.

#6about 2 minutes

Implementing burn rate alerting for proactive monitoring

Burn rate alerting uses multiple time windows to track how quickly the error budget is consumed, providing early warnings before an SLO is breached.

#7about 4 minutes

A practical demo of creating SLOs and dashboards

This demonstration walks through creating a transform, defining a custom SLO with a burn rate alert, and visualizing the results in a Kibana dashboard.

#8about 2 minutes

Managing SLOs and transforms with Elasticsearch APIs

Use the Dev Tools console in Kibana to interact with APIs for programmatically managing SLOs and transforms, which is useful for automation and troubleshooting.

#9about 3 minutes

Fostering cross-team collaboration with SLOs

Implementing SLOs is an organizational effort that requires collaboration between SRE, developers, product, and support teams to align on reliability goals.

#10about 5 minutes

Actionable takeaways for SREs on incident management

SREs should focus on understanding user expectations, defining key SLIs, balancing reliability and innovation, and using historical data to inform decisions.

#11about 9 minutes

Q&A on setting realistic SLOs and choosing tools

The discussion covers how to set achievable SLO targets, handle variance in serverless environments, and the benefits of in-house versus open-source monitoring tools.

#12about 7 minutes

Q&A on team alignment and prioritizing SLO work

This section addresses how to align multiple teams on SLOs through regular communication and how to prioritize improvements in the backlog, often driven by incidents or user feedback.

#13about 10 minutes

Q&A on third-party dependencies and new features

The final questions cover strategies for managing SLOs tied to third-party dependencies and balancing the introduction of new features with service reliability.