Shulou(Shulou.com)05/31 Report--

This article will explain in detail how to analyze OpenTelemetry and TSW in Trace. The content of the article is of high quality, so the editor will share it with you for reference. I hope you will have some understanding of the relevant knowledge after reading this article.

OpenTracing is the latest open distributed tracking standard for applications and OSS packages. Developers with experience of building microservices on a large scale understand the role and importance of distributed tracking: logging and metrics monitoring for each process has its place, but none of them can reconstruct the complex journey that transactions go through as they propagate across the distributed system.

The Origin of Trace

Trace is not a new concept, with the beginning of software commercialization, how to quickly locate the production environment is the direction that engineers continue to explore. Most of the early software architectures are CS architecture (Client and Server). In order to provide a quick and convenient way for engineers to solve problems, each language has been working hard on Log,Debug and Trace.

PrintStackTrace (), the most familiar Java code, is the standard tracking logic that makes the call relationships and order clear and clear. With the popularity of SOA (Service Oriented Architecture), standardized interaction mechanisms between services are gradually established, but most applications are still based on a single structure, so it does not bring great changes to tracking and troubleshooting.

The rise of the Internet and the geometric increase of traffic, the decoupling between services, the independent expansibility of modules and the changes in the organizational structure of the business team have essentially changed the software architecture. Break up the original single applications, micro-services almost become a rigid demand. The architecture of microservices disrupts the original link tracking and troubleshooting logic for simple monolithic development.

In this context, distributed link tracking has become a strong demand for business monitoring and fault location in most companies. This leads to the topic we want to discuss below, OpenTelemetry, which is also one of the most popular link tracking projects in the industry.

The past Life and present Life of OpenTelemetry

To tell the story of OpenTelemetry, we also need to review some important events of the past five years.

On Christmas Day, December 25, 2016, OpenTracing released version 1.0, which provides a Tracing-oriented specification for compatible platform-neutral API between microservice invocations. Over the next two years, various external components, languages and frameworks have continuously contributed to the ecology of OpenTracing, and OpenTracing has become a project of CNCF incubating.

A project that seemed so consistent with mainstream needs was challenged by Google in 2018. On January 17th, 2018, Google released OpenCencus (Microsoft joined in June 2018) as a vendor-independent performance acquisition and link tracking platform for microservice architecture.

In terms of positioning, OpenCencus not only provides Tracing standards and API specifications, but also integrates the construction of performance collection, and the goal is to create a solution that integrates performance monitoring and link tracking. Ecologically, OpenCensus directly provides the support of SDK corresponding to most languages and Framework that is widely used. From then on, the industry began round after round of discussion about whether to use OpenTracing or OpenCensus.

Finally, on May 21, 2019, an official announcement from Google came that OpenCensus and OpenTracing had been incorporated into the new CNCF project, which is called OpenTelemetry. On the other hand, OpenTelemetry not only does Tracing,Metrics, but also implements a similar neutral solution for Logging, which basically eats the upstream, middle and lower reaches of the troubleshooting process and strives to create a comprehensive specification for troubleshooting positioning.

At present, OpenTelemetry is a SandBox project under CNCF, and several more distinctive projects under SandBox include Chaos Mesh,Cert Manager and K3s. OpenTelemetry is fully compatible with OpenTracing and OpenCensus, will produce more comprehensive specifications, and many specifications will contribute to W3C-like.

At present, some of the hottest link tracking communities in the industry, such as Apache SkyWalking,Zipkin,Jaeger, are developed based on OpenTracing, and there is a trend of gradually expanding to OpenTelemetry. These open source builds usually start with client access, provide both data collection solutions (Data Collector) and data processing solutions (Data Aggregator), and finally pour data into other open source builds, such as Elasticsearch and Cassandra. For most of the customers who need to use it, self-building is actually a troublesome thing. For example, the example architecture diagram provided by Jaeger is as follows:

Figure 1

In order to achieve an ideal effect, there are too many things that customers need to build themselves, Kafka/ Flink/ DB/ Jaeger. So in most scenarios, customers want cloud vendors to provide one-stop service, rather than let customers assemble the whole solution themselves.

Let's take an example of TSW (Tencent Service Watcher) construction provided by Tencent Cloud to see how large cloud manufacturers help customers access OpenTracing and OpenTelemetry ecology and solutions.

OpenTracing-based Cloud Link tracking Infrastructure TSW

TSW is the abbreviation of Tencent Service Watcher. It is a distributed full link tracking solution provided by Tencent Cloud. TSW is designed to embrace open source and provide full-link problem-solving capabilities.

Here, let's talk about embracing open source. OpenTelemtry is still in the SandBox project stage, so TSW's current data protocol is based on OpenTracing, and the client that is fully compatible with Apache Skywalking,Zipkin and Jaeger reports, which provides great convenience for the client Tracing of the micro-service system to report and select the model.

With the gradual maturity of OpenTelemetry, data types and functions that are fully compatible with OpenTelemetry have been scheduled for TSW. This design provides the greatest convenience for companies wandering at the crossroads without having to consider the future compatibility of the client. The fully self-developed back-end system, drawing lessons from the design of Jaeger and Skywalking, uses the mechanism of computing storage separation and multi-layer query to provide excellent performance output.

Figure 2

In addition to providing a flexible view of the link topology and the query analysis mechanism of the calling link, TSW also makes a deep aggregation of the upstream and downstream relationship of the service invocation, which can more intuitively segment and compare the success rate and response of the same call under different links.

Figure 3

Figure 4.

In order to provide the R & D team with the greatest convenience for troubleshooting, TSW is deeply integrating with cloud monitoring and logging services, and will provide an one-stop troubleshooting experience in the future, avoiding the trouble of opening multiple pages in one problem and constantly copying and pasting keywords in search, and aims to create an one-stop workstation for troubleshooting on the cloud.

On how to analyze the OpenTelemetry and TSW in Trace to share here, I hope that the above content can be of some help to you, can learn more knowledge. If you think the article is good, you can share it for more people to see.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.