With the expansion of Internet architecture, distributed systems become increasingly complex, and more and more components begin to be distributed, such as microservices, message sending and receiving, distributed database, distributed cache, distributed object storage, and cross-domain call. These components together constitute a complex distributed network.
In a survey of more than 800 developers, when asked “What are the three biggest challenges you face in building a highly available distributed system today?” , 57% of developers chose full link tracing.
On June 12, Ali Cloud launched link Tracing service Tracing Analysis, which provides full-link Tracing capabilities for distributed systems to help customers quickly find and locate various performance bottlenecks in distributed systems at the cost of 1/5 or less of self-built link Tracing systems.
Although distributed application architecture under microservice architecture meets the requirement of application horizontal expansion, how to diagnose distributed application becomes a challenge. Although there are open source solutions related to link tracing in the industry, there are challenges such as high r&d investment, high self-built cost, high technical risk, and difficult operation and maintenance.
Link Tracing Tracing Analysis is derived from EagleEye, which has been proven in large-scale practice within Alibaba. Based on The Opentracing standard, it is fully compatible with the open source community. It can realize the hosting of open source solutions such as Jaeger, Zipkin and Skywalking on Ali Cloud, and customers do not need to build infrastructure, saving operation and maintenance investment and technical risks. In addition, multi-language clients can report application link data to the link tracing console for link tracing.
Link Tracing Tracing Analysis can be used to analyze link topology, discover and locate slow requests, abnormal requests, and traffic anomalies, and collect statistics based on service tags. Taking an education industry customer as an example, link Tracing Tracing Analysis helped customers reduce the number of abnormal requests from 3% to 0.1% and troubleshoot more than five online problems.
In addition, link Tracing Analysis can help users collect distributed call information of all distributed microservice applications and related PaaS products, and view application dependency paths for business Analysis and stability assessment. Taking a financial industry customer as an example, link Tracing Analysis helped customers reduce the average application response time from 2 seconds to 500 milliseconds.
It is worth noting that link Tracing Tracing Analysis saves the local storage cost of a customer’s self-built infrastructure and only collects storage fees through cloud log storage. The overall machine cost is 1/5 or less than that of a self-built full link Tracing system, and provides a free usage allowance of 1000 requests per day.
Currently, Ali Cloud Link Tracing Analysis has been applied to finance, gaming, education, retail, artificial intelligence and other industries to help developers efficiently analyze and diagnose performance bottlenecks under distributed application architecture.
Q&A:
Q1: Can you pull link-tracing data through the API? A1: Yes. Collected links can be obtained through OpenAPI, displayed on the link tracing page, or viewed in the log service.
Q2: Non-Alibaba cloud services that can access link tracking? A2: Link tracing is open. As long as the customer’s application can access the public network, it does not matter whether it is deployed on the Ari Cloud.
Q3: Is there a correlation analysis on the impact of buried points on performance? A3: Buried point data is reported asynchronously and in batches, which has a limited impact on performance, generally around 1%. It depends on the number of buried points, and a large number of buried points has a greater impact. According to the current pressure data, the impact on performance is relatively small.
The original link
This article is the original content of the cloud habitat community, shall not be reproduced without permission.