SkyWalking vs Pinpoint: Which Open‑Source APM Wins for Your Stack?

This comparison is based on SkyWalking 6.0.0‑GA and Pinpoint 1.8.2 (latest as of 2019‑02‑19). Both open‑source APM products achieve non‑intrusive bytecode injection, unlike Zipkin or CAT, which were rejected for their code‑level intrusion.

Community Comparison

SkyWalking has a more active community, being an Apache incubating project with a Chinese community and direct communication channels (e.g., Apache SkyWalking group 392443393). Pinpoint, developed in Korea, has a language barrier and comparable recent commit frequency.

Supported Languages

Pinpoint supports Java and PHP only. SkyWalking supports Java, C#, PHP, and Node.js, and offers lower development cost for custom probes (e.g., Go).

Note: A Node.js agent for Pinpoint existed on GitHub (peaksnail/pinpoint-node-agent) but is no longer maintained.

Protocol Comparison

SkyWalking uses gRPC (http deprecated in 6.x) while Pinpoint uses Thrift. Neither protocol is inherently superior.

Storage Comparison (Important)

Pinpoint only supports HBase, requiring a dedicated HBase cluster and offering limited query capabilities (row‑key get or scan). SkyWalking supports multiple storages, primarily Elasticsearch in production, and allows flexible query dimensions (time, service, trace status, path, trace ID, etc.). Both support TTL; SkyWalking configures it in application.yml, Pinpoint via HBase scripts (e.g., TTL => 5184000 for ApplicationTraceIndex, TTL => 15552000 for SqlMetaData_Ver2).

ES is not universally superior to HBase: ES excels at search, HBase at storage. Choose based on log volume and query requirements.

UI Comparison

Pinpoint’s UI is slightly better, especially for topology visualization. A custom SkyWalking UI (rocketbot) is available on GitHub.

Extensibility Comparison

SkyWalking is designed to be pluggable. Custom storage can be added by implementing

org.apache.skywalking.oap.server.library.module.ModuleProvider

. Pinpoint lacks such extensibility. SkyWalking also supports OpenTracing and community‑maintained plugins for C#, PHP, and Node.js, while Pinpoint’s plugin ecosystem is more limited.

Both support plugin development; Pinpoint’s guide is at http://naver.github.io/pinpoint/1.8.2/plugindevguide.html, SkyWalking’s at

https://github.com/apache/incubator-skywalking/blob/master/docs/en/guides/Java-Plugin-Development-Guide.md

.

Alarm Comparison

Pinpoint requires MySQL to persist alarm rules and users, adding maintenance overhead. SkyWalking configures alarms directly in config/alarm-settings.xml and supports webhook notifications, custom silence periods, and a richer set of service‑level indicators.

Pinpoint alarm reference: http://naver.github.io/pinpoint/1.8.2/alarm.html

SkyWalking alarm reference: https://github.com/apache/incubator-skywalking/blob/master/docs/en/setup/backend/backend-alarm.md

JVM Monitoring

SkyWalking monitors Heap, Non‑Heap, and both YGC and FGC. Pinpoint monitors Heap, Non‑Heap, FGC, DirectBufferMemory, and MappedBufferMemory, but lacks YGC.

Service Monitoring

Pinpoint provides CPU, open file descriptors, datasource, active threads, response time, and TPS. SkyWalking provides CPU, SLA, response time, and CPM. Overall, they are comparable.

Trace Granularity

Pinpoint offers finer granularity, showing detailed parameters for each traced interface.

Filter Trace

Both support ant‑style filter expressions. Pinpoint configures filters via a web UI wizard; SkyWalking uses the apm-trace-ignore-plugin for flexible rule definitions.

Performance Overhead

Pinpoint’s detailed data collection incurs higher overhead, while SkyWalking’s conservative default strategy results in lower impact. A stress test (image source: Juejin ) confirms this.

Release Package

SkyWalking ships as a set of JARs with a simple start.sh script. Pinpoint’s collector and web components are packaged as WAR files, requiring a servlet container (e.g., Tomcat).

Component Support Comparison

Web Containers : Pinpoint supports most containers, including commercial ones; SkyWalking only supports open‑source containers.

RPC Frameworks : SkyWalking supports a broader range, including Motan and SofaRPC.

Message Queues : SkyWalking adds RocketMQ support.

RDBMS/NoSQL : Pinpoint slightly ahead (supports MSSQL, MariaDB, HBase); SkyWalking lacks MSSQL, MariaDB, Cassandra, HBase.

Redis Clients : SkyWalking supports Jedis, Redisson, Lettuce; Pinpoint only Jedis and Lettuce.

Logging : Pinpoint lacks native log4j2 support (community plugin exists).

Overall, the choice depends on specific requirements: massive log storage favors Pinpoint (HBase), while multi‑language support, Elasticsearch search capability, and extensibility favor SkyWalking.

Conclusion

Both APM tools are excellent. Pinpoint excels in trace granularity, UI richness, and HBase‑based storage for massive logs. SkyWalking shines with an active Chinese community, broader language support, Elasticsearch search, pluggable architecture, and simpler deployment.

If you need massive log storage, consider Pinpoint.

If you value easy extensibility and multi‑language agents, choose SkyWalking.