A large-scale complete system consists of hundreds of distributed service systems. Each request routed in will pass through multiple business systems and leave footprints, and generate access to various DBs and caches, but these decentralized systems For troubleshooting, or process optimization are limited. For such cross-process scenarios, it is particularly important to collect and analyze massive logs first.
To track the complete call chain of each request, collect performance data of each service on the call chain, calculate performance data and compare performance indicators (SLA), and even feed back into service governance in the far future, then this It is the goal of service governance.
In the industry, Taobao's Eagle Eye and Twitter's zipkin are similar systems, both of which originated from Google's dapper.
To sort it out, Google is called Dapper, Taobao is called Eagle Eye, Twitter is called ZipKin, Jingdong Mall is called Hydra, eBay is called Centralized Activity Logging (CAL), Dianping is called CAT, and we are called Tracing.
Distributed call chain tracing systems typically have several design goals:
1. Low intrusion, as a non-business component, it should intrude into business systems as little as possible, be transparent to users, and reduce the burden on developers.
2. Flexible application strategies that can determine the scope and intensity of the data collected.
3. From the collection and generation of data, to the calculation and processing of data, to the final presentation, it must be as fast as possible.
4. Decision support, does this data play a role in decision support, especially from a DevOps perspective.
5. Visualization is the top priority.