Tencent Cloud Application Performance Management

Tencent Cloud Application Performance Management is a cloud APM service for teams operating distributed applications and microservices. It is strongest when engineers need request-level evidence, dependency maps, and correlated metrics rather than isolated server charts. The usage-based model and free allowance make evaluation accessible, while retention and sampling limits deserve close attention before production adoption.

Overview

APM brings application metrics, traces, and logs into one analysis workflow. It follows requests across services, exposes failures and exceptions, and relates those events to application dependencies. Dimension-based aggregation by API, exception, and database call helps identify slow endpoints, expensive SQL statements, and recurring faults.

Real-time monitoring uses probes without requiring code modifications, which suits teams that cannot easily alter application binaries. Although the deployment is cloud-based, applications outside Tencent Cloud can connect through a public endpoint or private connectivity such as Direct Connect to a Tencent Cloud VPC. That supports mixed environments, provided the required network path is available.

Key features

Tracing and service maps

APM reconstructs request paths across microservices and captures request parameters, transaction data, errors, exceptions, method stacks, and instance-environment information. Its global application topology view discovers dependencies and helps locate bottlenecks at the application or component level. The combination is valuable during incidents where a visible outage originates in a downstream service.

Metrics, logs, and databases

Performance and exception data can be grouped by operational dimensions, giving teams a practical route from a symptom to a specific API, database call, or exception pattern. Named database integrations include MySQL, Oracle, MongoDB, and Redis. Traces can be associated with Tencent Cloud Log Service logs, while Cloud Monitor dashboards provide another view of service health. Trace delivery to CKafka is supported when configuration and permissions are set.

Instrumentation and compatibility

OpenTelemetry support provides a current instrumentation path, and compatibility with SkyWalking and Jaeger tracing protocols can simplify migration by changing reporting configuration rather than rebuilding an observability pipeline. Tencent Cloud also publishes APIs for managing business systems and applications and for querying topology, metrics, and traces, which suits teams integrating monitoring into their own tooling.

Access control

Cloud Access Management policies offer full read/write, read-only, and business-system-level permissions. That separation is useful when developers, operators, and application owners need different levels of access to production telemetry.

Pricing

APM uses freemium, usage-based pricing. New users receive a 15-day trial capped at 100 million reported Spans and seven days of trace storage. Pay-as-you-go also includes one million Spans per day free for reporting and one million Spans per day free for retention; beyond that, reporting is $0.02 per million Spans per month, billed hourly, and retention is $0.012 per million Spans per day. Retention duration is selectable, so this option fits teams that want usage to track workload and need control over storage duration.

Prepaid packages provide three days of trace storage and use fixed 10% sampling while retaining exceptional traces. That lower retention window and sampling policy are a material trade-off for investigations requiring older or complete traces.

• Developer Experience Package: custom pricing, three days of trace storage, and 3,600 Agent*Hours. Suits a small development workload.
• Developer Standard Package: custom pricing, three days of storage, and 28,800 Agent*Hours for a larger development footprint.
• Enterprise Basic Package: custom pricing, three days of storage, and 273,600 Agent*Hours for established production usage.
• Enterprise Professional Package: custom pricing, three days of storage, and 1,080,000 Agent*Hours for higher-volume estates.
• Flagship Package: custom pricing, three days of storage, and 3,600,000 Agent*Hours for the largest listed capacity.
• Developer - Experience Edition: $150 a month, billed for one month, with 3,600 Agent-hours, three-day storage, and 10% sampling with exceptional traces retained. It is a straightforward prepaid entry point, but not a long-history archive.

Platforms

APM is delivered as a cloud service and supports API, Linux, self-hosted, and web environments. OpenTelemetry, SkyWalking, and Jaeger compatibility broadens integration choices, while external applications still require network access to Tencent Cloud APM.

Who it's for

Choose APM for microservice teams that need distributed tracing, topology discovery, database-aware diagnosis, and Tencent Cloud integrations in one service. It is also a sensible option for organizations standardizing on OpenTelemetry or migrating from SkyWalking or Jaeger. Teams needing lengthy trace history, complete unsampled retention, or disconnected operation should look beyond the prepaid model and assess alternatives.

Pros and cons

• Pro: Request paths, topology, metrics, logs, and exception analysis connect symptoms to likely bottlenecks.
• Pro: Probe-based monitoring without code modifications can shorten instrumentation work.
• Pro: OpenTelemetry plus SkyWalking and Jaeger compatibility supports migration and mixed tooling.
• Pro: A 15-day trial and daily free Span allowances reduce the cost of initial evaluation.
• Con: Prepaid packages retain traces for only three days and sample at a fixed 10%, limiting historical and complete investigations.
• Con: Usage charges rise with reported Spans and retained data, so high-volume systems need careful budgeting.
• Con: External deployments depend on public or private network connectivity to Tencent Cloud.

Alternatives

Distributed Tracing Tools is the relevant category when you want to compare APM and tracing products broadly. Data Lake Software is a better starting point for storage and analytics architecture, while Translation Software serves a different requirement.

HPCC Systems is a free, Apache 2.0-licensed self-hosted option when an open-source data platform matters more than managed APM views. Parseable suits teams wanting free OSS log querying, dashboards, threshold alerts, OIDC SSO, unlimited users, and customizable retention or object storage.

Apache Hadoop HDFS and Apache Ozone are free open-source storage choices for teams prioritizing self-hosted data infrastructure. AWS Lake Formation fits AWS-centered permission and integrated-service workflows, with permissions provided at no charge while integrated services use their standard rates. Cloudera Data Lake Service is a paid, contact-sales option. HPE Ezmeral Data Fabric is a paid platform with reseller-dependent pricing and a Canada listing starting at C$500.00. lakeFS is worth considering for managed lakeFS Cloud or self-managed, on-premises, own-cloud, or air-gapped deployments.

Verdict

Tencent Cloud APM is a strong fit for distributed-application teams that want integrated traces, service maps, metrics, logs, and Tencent Cloud connectivity with OpenTelemetry support. Start with the trial or pay-as-you-go allowance when workload volume is uncertain. Look elsewhere if three-day prepaid retention, fixed 10% sampling, or mandatory network access conflicts with your investigation and deployment requirements.