Hi Tianjiang, 1. I really like the "merge (statistics)", can it be a tool/remote call of IoTDB?
2. the data size is reduced from "32G + 9.3G" to "26G". Is that because the compression is improved or the chunks (and their metadata) are reduced? (If the latter, what the threshold of the number of data points in a chunk for merging is?) 3. why you can attach a file on the mail list... Best, ----------------------------------- Xiangdong Huang School of Software, Tsinghua University 黄向东 清华大学 软件学院 Tian Jiang <[email protected]> 于2019年9月2日周一 上午2:15写道： > > Hi, > > I have done a simple experiment comparing enabling merge or not and here > are some basic conclusions: > > 1. Compared with disabling merge, enabling merge does not affect insertion > significantly. > 2. After merges are done, cold cache query (performed on a newly rebooted > IoTDB) is significantly improved. > 3. After merges are done, hot cache query (performed on a already > queried IoTDB) is not affected. > 4. With background insertions, merges can be very slow and without them, > merges are a little faster than insertions. > 5. When merges are done, the total size of data files reduces > significantly. > > For detailed experiment results, please see the attachment. > > Tian Jiang >
