Hi all,

I noticed that there are some users want to use IoTDB in their
applications, and there are still a lot of things needed to do for IoTDB.
To build an excellent Time Series Database, I think we can try to add the
following features into IoTDB:

* Distributed version.
  ** Plan: we hope to finishing developing it in the first half year of
2019. A beta version will be proposed in the middle of 2019.

* System monitoring. A module to monitor the active threads, memory cost,
read and write latency, etc..

* Multi-resolution (also called Roll-up). As time series data may be
collection in a high frequency, this function can roll data up
automatically.  That is, we not only stores the original data, but also
data with second-granularity, minute-granularity, hour-granularity and
day-granularity.

* Update, Deletion and TTL. Support update or delete data when given a time
range. Support Time-To-Live.

* Segmentation. A time series can be segmented into several sub time
series. For example, a time series represents the heartbeat of a person,
after segmentation, we can get several sub time series, one is for
describing the behavior of sleeping in the night, one is for describing
walking for work, etc..

* Restful API. A native restful API (for writing and query) will help more
applications using IoTDB.

* Index. An index framework will be provided and some time series oriented
index (such as KV-index) will be added.

* Time-Frequency domain transfer.  (a design document is here:
https://github.com/thulab/iotdb/issues/499 )

I know there are many more other interesting functions in time series
domain.
Please add your viewpoint here, and we can discuss in detail, such as the
priority, the function definition, the implementation, and the time
schedule.

Best,
-----------------------------------
Xiangdong Huang
School of Software, Tsinghua University

 黄向东
清华大学 软件学院

Reply via email to