Network Security Internet Technology Development Database Servers Mobile Phone Android Software Apple Software Computer Software News IT Information

In addition to Weibo, there is also WeChat

Please pay attention

WeChat public account

Shulou

What is the new way for eBa to manage a large service architecture

2025-01-25 Update From: SLTechnology News&Howtos shulou NAV: SLTechnology News&Howtos > Internet Technology >

Share

Shulou(Shulou.com)06/02 Report--

What is the new way for eBa to manage the huge service architecture? I believe many inexperienced people are at a loss about it. Therefore, this article summarizes the causes and solutions of the problem. Through this article, I hope you can solve this problem.

Today we share an article that explains how eBay uses a knowledge graph to manage a large service architecture.

Managing and understanding a large ecosystem in a service architecture is no small challenge: imagine that there are more than 3000 service application clusters in its production system, especially for eBay. Each application develops independently and has different functions and development methods. The lack of documentation and the lack of a correct understanding of internal customers may hinder efficient development.

EBay's vision (known as INAR, smart architecture) is to build a sustainable service architecture by providing automated visibility, evaluation, and governance intelligence (Governance Intelligence). For this reason, we have developed a new method to model and deal with the application ecosystem using knowledge graph.

Governance Intelligence,GI, proposed by Dr. Francis Ohanyido, mainly refers to the use of information and communication technology applications and technologies to identify, extract, analyze and describe governance data.

Knowledge graph is a commonly used term, but its exact definition has always been controversial. Basically, knowledge graph is a programmable way to model the knowledge domain using subject content experts, interrelated data and machine learning algorithms. For eBay, the application / infrastructure knowledge graph is a heterogeneous attribute graph, which can improve the visibility of the architecture, operational efficiency and developer productivity, and ultimately enable customers to get a better experience when visiting the website.

This article explains how the knowledge graph of the eBay architecture is developed, what benefits eBay has gained from it, and the use cases we will see in this approach now and in the future.

Three major challenges

The vision of a smart architecture aims to address three key challenges of a service architecture:

Blindness: architectural issues can be difficult to observe, such as inappropriate dependencies between software and / or hardware, or imagine the infrastructure and ecosystem of eBay, as well as custom search. This is a problem because popular software and services often evolve and tend to be single, resulting in service redundancy and duplication of functions.

Ignorance: the lack of testability of service architecture or technical debt (the extra rework required to take a simpler upfront approach is worse in the long run) may hinder you from developing the metrics you need to improve operational efficiency. As the management master Peter Drucker famously said, "without measurement, there is no management." If you can't measu re it, you can't improve it.

Originality: diagnostics, engineering, and runtime automation do not exist. Therefore, artificial intelligence can not be applied to IT operations, so it is difficult to find anomalies in operations.

It is clear that a clearer understanding of our ecosystem is needed if we are to meet the needs of 183 million buyers. Our goal is to provide better visibility, provide pattern / anomaly detection, and automate and enhance IT operations. This gives us the idea of using knowledge / attribute diagrams.

Building links: behavioral metrics and intelligent layering

The diagram is built using real-time metrics, business characteristics, and operational metadata. Ultimately, the goal of the diagram is to link data sources and break the boundaries between isolated administrative domains. The following is a higher-level description:

The first step in developing a knowledge graph is to calculate the best application metrics and apply machine learning algorithms to cluster applications automatically. We have developed metrics to measure application popularity based on real-time traffic and runtime dependencies.

We calculated the metrics for all eBay clusters and clustered all services using techniques called K-average (K-means) and Canopy clustering, and then based on their popularity scores. This enables us to group ecosystems into different categories, such as their level of activity. We found that 77% of the clusters were marked as low activity.

Visible understanding: Graph search and result Visualization

One of the goals of using knowledge graphs is to improve the productivity of developers so that they can retrieve the information they need more effectively. Currently, developers must accept the information they need through a number of tools.

In order to improve production efficiency, we build a complete batch processing system to obtain data from different sources and construct knowledge graph automatically. We also build intelligent graph search, which dynamically generates a query to explore the knowledge graph, including service measurement and intelligent layering. The following data schema is designed at the application (pool) level, with bold or black borders, and will be enabled as the first "baby" step:

We have a better understanding of the ecosystem by connecting cloud native data, hardware, people, code, and business. Visualization provides a wealth of information in a way that can be understood and operated quickly. In the following service dependency example: we randomly select 18 services and visualize them through one of the default methods. In the following illustration, the border thickness represents the attribute (volume) of the edge, and the node size represents the behavior metric. Different colors represent a team or organization (for example, yellow represents a domain team).

The eBay dependency system "Galaxies" uses a proof of concept (Proof of Concept,POC). Now, the graph schema is extended as follows:

Identify low efficiency

We calculated metrics and intelligent services in more than 3000 eBay production clusters. Three senior architects manually verified the initial results of popularity metrics and automatic clustering.

The results are surprising, but they also provide a lot of information. About 10% of highly active applications run in inaccurate availability zones, which can affect operational performance and uptime.

For eBay, knowledge graph has become an important tool (Galaxies), which enables us to provide customizable visualization, application metrics, intelligent layering and graph search.

The system provides top-down and bottom-up views of applications, as well as dependencies and higher accuracy; enriching data to enhance application compliance; governance with clear ownership details; and recommendations for operational performance.

Next, we plan to enhance the diagram by displaying all the causality details of each event on the diagram to support site anomaly detection (this is an initial effort).

We also plan to extend this diagram to include service API metadata, which will make service layering, recommendation, and clustering possible. The knowledge graph is expected to be a key tool for understanding our ecosystem and meeting our customers' expectations for faster and better service.

After reading the above, have you mastered eBa's new approach to managing a large service architecture? If you want to learn more skills or want to know more about it, you are welcome to follow the industry information channel, thank you for reading!

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.

Views: 280

*The comments in the above article only represent the author's personal views and do not represent the views and positions of this website. If you have more insights, please feel free to contribute and share.

Share To

Internet Technology

Wechat

© 2024 shulou.com SLNews company. All rights reserved.

12
Report