SLOs and Error Budgets in Production: Lessons and Pitfalls
TL;DR
A complete, up-to-date breakdown of slos for developers and founders. It covers the core ideas, the trade-offs that matter, a practical workflow, real numbers, and the questions people ask most — written to be skimmed, applied, and shared.
Key takeaways
- Use traces to answer 'where is the time going in this request,' metrics to answer 'is the system healthy at scale,' and logs to answer 'what exactly happened here.'
- Treat the error budget as a shared currency: when it is healthy you ship features, when it is exhausted you freeze and fix reliability.
- Adopt structured, correlated logs (with trace and span IDs) so you can pivot from a symptom to the exact request path that caused it.
- Watch cardinality on metric labels - a single unbounded label like user_id or request_id can explode a Prometheus time series database.
- Run blameless postmortems and feed their action items back into your alerting, SLOs, and automation to shrink the next incident.
This is a practical, up-to-date guide to Slos — what it is, why it matters in 2026, and how to apply it in real projects. It is written for developers and founders who want clear answers and proven best practices, not filler.
Whether you're just starting out or leveling up, treat this as a working reference you can return to. Every section is built to be skimmed, applied, and shared.
Controlling cost and cardinality
Observability data frequently grows faster than the systems it watches, and unmanaged telemetry can become one of the larger lines on a cloud bill, so cost control is now a first-class engineering concern. The dominant driver for metrics is cardinality - the number of unique label combinations - because attaching an unbounded value like a user ID or full URL to a metric can create millions of time series and overwhelm a database. For logs and traces, sampling is the primary lever: head-based sampling decides up front, while tail-based sampling in the OpenTelemetry Collector keeps the traces that are actually interesting, such as slow or errored requests. Tiered storage strategies move older or lower-value data to cheaper object storage, and tools increasingly let teams aggregate or drop low-signal data at the Collector before it ever reaches a paid backend. The guiding principle is to retain high-context data about anomalies and aggregate the routine, rather than storing everything at full fidelity forever.
How OpenTelemetry unifies instrumentation
OpenTelemetry (often abbreviated OTel) is a CNCF project that provides a single, vendor-neutral set of APIs, SDKs, and wire protocols for generating metrics, logs, and traces. It emerged from the merger of the earlier OpenTracing and OpenCensus projects, which ended a period of fragmentation where instrumenting for one vendor locked you out of others. The core payoff is portability: you instrument your code once against the OTel API, export data over the OpenTelemetry Protocol (OTLP), and can then send it to Prometheus, Jaeger, Grafana, Datadog, Honeycomb, or any compatible backend without touching application code again. OTel also defines semantic conventions - standardized names for common attributes like http.request.method or db.system - so telemetry from different languages and libraries is consistent and joinable. Auto-instrumentation agents exist for languages like Java, Python, .NET, and Node.js, letting teams capture rich traces with little or no manual code.
Prometheus and the metrics ecosystem
Prometheus is an open-source monitoring system and time series database that pioneered a pull-based model, scraping metrics from HTTP endpoints that applications expose in a simple text format. Its dimensional data model, where each time series is identified by a metric name plus a set of key-value labels, combined with the PromQL query language, made flexible slicing and alerting the norm in cloud-native operations. Prometheus is the de facto standard for Kubernetes monitoring, and its exposition format was formalized into OpenMetrics and is natively understood across the ecosystem. Because a single Prometheus server is designed to be simple and reliable rather than infinitely scalable, long-term storage and global querying are handled by projects such as Thanos, Cortex, Grafana Mimir, and VictoriaMetrics. Alertmanager, a companion component, handles deduplication, grouping, silencing, and routing of alerts to destinations like PagerDuty, Slack, or email.
What observability actually means
Observability is a property of a system that describes how well you can understand its internal state from the outputs it emits, a concept borrowed from control theory and adapted to software. In practice it means instrumenting applications and infrastructure so that when something goes wrong, you can ask new questions about behavior you did not anticipate in advance, rather than only checking pre-built dashboards. This is the key distinction from traditional monitoring, which excels at answering known questions about known failure modes but struggles with novel, emergent problems in distributed systems. Modern observability is usually discussed in terms of three primary signal types - metrics, logs, and traces - increasingly joined by continuous profiling. The goal is not to collect everything, but to collect the right high-cardinality, high-context telemetry so that unknown-unknowns become debuggable.
AIOps and anomaly detection
AIOps refers to applying machine learning and statistical analysis to operations data to reduce noise, surface anomalies, and speed up root-cause analysis at a scale humans cannot manually monitor. Common applications include alert correlation and deduplication (grouping a storm of related alerts into a single incident), dynamic baselining that learns normal traffic patterns instead of relying on static thresholds, and automated anomaly detection on high-dimensional metrics. Vendors such as Datadog, Dynatrace, New Relic, and Splunk market AIOps capabilities, and the newest wave layers large language models on top to summarize incidents, draft postmortems, and suggest likely causes from correlated telemetry. The value is real when it cuts through alert fatigue and shortens investigation time, but practitioners caution that opaque models can erode trust if they cannot explain why they flagged something. The pragmatic stance going into 2026 is to use AIOps to augment on-call engineers - triaging and summarizing - rather than to fully automate judgment.
Distributed tracing in microservices
Distributed tracing addresses a problem that metrics and logs alone cannot: understanding a single request as it fans out across dozens of independent services, queues, and databases. Each unit of work becomes a span with a start time, duration, status, and attributes, and spans are linked through a shared trace context that is propagated across network calls via standardized headers like W3C Trace Context. The result is a waterfall view showing exactly which service or dependency added latency or threw an error, which is invaluable for debugging tail latency and cascading failures. Popular open-source backends include Jaeger and Grafana Tempo, and OpenTelemetry has become the standard way to generate the spans that feed them. Because tracing every request at high volume is expensive, teams rely on head-based or tail-based sampling to keep representative and interesting traces while controlling cost.
Slos: Key Facts and Data
According to recent industry research and the official documentation linked below:
- Google popularized the SRE discipline through its 2016 book 'Site Reliability Engineering,' and the model of running services against explicit SLOs and error budgets has since been adopted well beyond Google.
- The DORA research program links elite software delivery performance to strong operational practices, and metrics like change failure rate and mean time to restore (MTTR) are commonly tracked alongside SLOs as of 2025.
- Industry surveys such as the CNCF annual survey indicate that Prometheus is one of the most widely adopted tools for metrics collection in cloud-native environments, with usage spanning a large majority of Kubernetes operators.
Quick-Reference Summary
A map of what this guide covers:
| Topic | What you'll learn |
|---|---|
| Controlling cost and cardinality | Observability data frequently grows faster than the systems it watches |
| How OpenTelemetry unifies instrumentation | OpenTelemetry (often abbreviated OTel) is a CNCF project that provides a single |
| Prometheus and the metrics ecosystem | Prometheus is an open-source monitoring system and time series database that pioneered a pull-based model |
| What observability actually means | Observability is a property of a system that describes how well you can understand its internal state from the outputs it emits |
| AIOps and anomaly detection | AIOps refers to applying machine learning and statistical analysis to operations data to reduce noise |
| Distributed tracing in microservices | Distributed tracing addresses a problem that metrics and logs alone cannot |
How to Get Started with Slos
A simple path that works:
- Learn the fundamentals of Slos from primary sources, not just tutorials.
- Build one small, real project end to end.
- Get feedback, refactor, and add tests.
- Ship it publicly and document what you learned.
- Repeat with a slightly harder project each time.
Build It with a World-Class Full Stack Developer
Sandeep Kumar Chaudhary is a full stack world-class developer. If you want to turn this into a real, production-ready product, get in touch — message directly on WhatsApp at +9779802348957 for a fast, no-pressure consult.
You can also explore the projects already shipped to thousands of users, or start a conversation here.
Final Thoughts
Use traces to answer 'where is the time going in this request,' metrics to answer 'is the system healthy at scale,' and logs to answer 'what exactly happened here.'. The developers and teams who win in 2026 pair strong fundamentals with consistent shipping. Start small, stay curious, build in public, and revisit this guide as your skills grow.
Sources and Further Reading
Frequently Asked Questions
What is slos?
OpenTelemetry (often abbreviated OTel) is a CNCF project that provides a single, vendor-neutral set of APIs, SDKs, and wire protocols for generating metrics, logs, and traces. It emerged from the merger of the earlier OpenTracing and OpenCensus projects, which ended a period of fragmentation where instrumenting for one vendor locked you out of others. This guide covers slos end to end — core concepts, best practices, concrete data, and a step-by-step approach you can apply right away.
What is a blameless postmortem?
A blameless postmortem is a written review after an incident that focuses on how the system, tooling, and processes allowed a failure rather than on which individual made a mistake. The premise is that people generally act reasonably given the information and tools they had, so punishing individuals hides the real systemic causes and discourages honest reporting. The output is a set of concrete, tracked action items to prevent recurrence, which is what turns an incident into lasting improvement.
What is the difference between an SLI, an SLO, and an SLA?
An SLI (Service Level Indicator) is a measured quantity such as the percentage of requests served under 300 milliseconds. An SLO (Service Level Objective) is your internal target for that indicator, for example that 99.9 percent of requests meet the latency threshold. An SLA (Service Level Agreement) is a contractual commitment to customers, usually looser than your internal SLO, with financial or legal consequences if you breach it.
What causes high cardinality and why is it a problem?
Cardinality is the number of unique combinations of a metric's labels, and it explodes when you attach unbounded or high-variety values such as user IDs, request IDs, email addresses, or full URLs as labels. Each unique combination becomes its own time series, so a single careless label can create millions of series and overwhelm the memory and storage of a system like Prometheus. The fix is to keep high-variety identifiers out of metric labels (put them in traces or logs instead) and reserve labels for bounded, low-variety dimensions like status code or region.
What is the difference between monitoring and observability?
Monitoring tells you whether known failure conditions are occurring by tracking predefined metrics and thresholds, answering questions you anticipated in advance. Observability is a broader property that lets you ask new, unanticipated questions about your system's internal state from its outputs, which matters most for novel problems in complex distributed systems. In short, monitoring is a subset of what a good observability practice enables; you still monitor, but you can also explore.
Sandeep Kumar Chaudhary
Full Stack Software Developer· Nepal's SEO, AEO, GEO & AIO expert and share-market educator. More about me