What Makes Lightstep Different
by David Klein
Lightstep’s unique Satellite Architecture analyzes 100% of unsampled transaction data from virtually any source — mobile apps, web clients, serverless, and more — offering end-to-end visibility into any trace, operation, or service.
Uncompromised by head or tail sampling, you have complete access to a full-fidelity distributed tracing dataset — available through a simple search bar, real-time diagrams of service dependencies, and context-rich visual summaries of patterns, outliers, and performance issues.
Since Lightstep adheres to OpenTracing standards (and helped to create the OpenTracing project, along with partners in the OSS ecosystem), the platform can support proxies, service meshes, and any language from C# to Go to Swift.
There is no limitation to the number or nature of the tags or log payloads that can be associated with a trace. This means you can navigate and explore any segment of your application, no matter how fine-grained the criteria, and review the performance of a host, user, geography, release version, experiment id, or even specific customer.
Lightstep automatically determines, and clearly highlights, the critical path of a given trace.
This allows you to quickly identify what’s causing an issue — whether it’s an n+1 query, a slow service, or something running in sequence that should be in parallel. In effect, you’re able to understand what not to work on, and to spend cycles on performance optimizations that are likely to drive a positive change in end user or customer experience.
Lightstep’s Service Diagrams offer an interactive, real-time, and dynamic overview of system performance and architecture.
Service Diagrams are unique in that they build dynamic system maps to answer specific questions about performance: highlighting bottlenecks in real time and reducing the search space for identifying where performance improvements should be made.
Through these clear, visual maps of any location in your system or stack, you can quickly manage issues both up and downstream, and more easily identify when — and where — third-party dependencies are slow or non-performant.
Ultimately, Service Diagrams show how services relate to one other, reveal dependencies, and highlight services that contribute to the latency of a transaction or are experiencing errors.
Lightstep enables near instantaneous sharing of any moment. This includes detailed views of performance for any process, operation, service, or system are automatically captured, recorded, and available in perpetuity via a unique url.
These Snapshots can be shared to quickly resolve issues during a live incident or better understand an event during a post mortem (rather than relying on the imperfect recollection of individual teammates).
It’s also the most immediate way to see — and explain to the rest of the team — what happened before or after a release.
Snapshots include detailed latency histograms, historical layers that provide temporal context, and hundreds of relevant example traces to help explain the symptoms observed — all of which durably persist for historical review.
You can even share a specific (possibly non-performant) span with your team, as these unique urls can be appended to include hyper-specific data, which is much easier to understand than sharing, say, a complete 1,000-line trace.
Lightstep automatically surfaces the strongest signals for the root cause of any event. Over any period of time, Lightstep can analyze statistical relationships between system attributes, and immediately reveal commonalities between high- and low-latency traces.
With these insights, instead of having to guess which teams might need to work on an incident — and sending dozens of employees to investigate an event over which they have no influence — you can now send just the right people.
This eliminates unnecessary disruption to teams who are ultimately not needed for incident resolution, but are involved due to a lack of situational clarity.
Lightstep is purpose-built for large, complex systems. Individual customers generate as much as half a trillion spans in a 24-hour period — with no performance degradation whatsoever.
And you have continuous access to every tracing data point in your system. If you’ve ever had a trace, it still exists today, and it can be viewed, aggregated, and visualized in context.
With our Service Health for Deployments feature, you're able to see any regressions as soon as you deploy.
Our latency histograms show a clear indicition of when and where the regression happened. This will give you a full-context view of how successful your deployment was and where an issue may be.