基于Hive的支付SDK日志分析系统的设计研究

摘要

UniPay支付SDK是联通沃商店为了给开发者提供手机话费和第三方支付能力而推出的一站式应用内统一支付插件,支付SDK日志记录海量的用户终端信息、用户使用手机应用的行为记录等数据.针对传统数据仓库难于满足海量日志数据存储及处理等问题,设计一种基于Hive的支付SDK日志分析系统.测试结果表明,使用Hadoop框架及Hive数据仓库对海量支付SDK日志进行存储和处理,能很好地满足业务需求,对手机应用的设备激活量、日活跃用户数(DAU)、分时日志量以及用户支付转化率等指标的分析结果,对应用开发者升级优化其应用及运营人员的营销策略调整具有重要的参考价值.%UniPay payment SDK is Unicom Wo store launched a one-stop application within the unified payment plug-in for developers to provide mobile phone calls and third-party payment capabilities.Payment SDK log records a large number of user terminal information, the use of mobile phone application behavior records and other data.Aiming at the problem that the traditional data warehouse can not satisfy the massive log data storage and processing, a Hive-based payment SDK log analysis system was designed.The test results show that using Hadoop framework and Hive data warehouse to store and process the massive payment SDK log, can meet the business demand well.The analysis result of user behavior in device activation, daily active user, quantity of time-log and the user payment conversion rate can provide important references for application developers to optimize and upgrade their mobile applications, and for operational personnel to adjust their marketing strategy.