This paper introduces a distributed log service System treating logs as data streams built with Node.js. and provides a solution to the problem that with the service number increases, the operation based on files gives rise to disadvantages that files I/O have performance bottleneck and have to be checked at multiple places. Logs are considered as a sort of data stream rather than files. Write a new log service system named octoLogStreamlet which collects the streams to a central storage with Node.js. Combining Node.js’ non-blocking I/O model and outstanding performance of NoSQL database, the system meets the requirements to cope with distributed data streams. The system is suitable for computer clusters needing centralized storage of logs, which provides great convenience for processing and utilizing logs.