Skip to content

T
TDengine

taosdata / TDengine
1 年多 前同步成功

通知 1185
Star 22016
Fork 4786
T
TDengine
未验证 提交 4311b0e5 编写于 作者: sangshuduo's avatar sangshuduo 提交者： GitHub
浏览文件
操作

Docs/sangshuduo/td 13307 add desc to index (#10326) 

* [TD-13307]<docs>: add description to docs index.

* fix the link of how to use taosBenchmark

* fix wrong links

* fix markdown format

* refine getting started section

* update

* fix install doc link

* fix chinese wording

* adjust section order

* adjust few format

* fix markdown format
上级 f7b580e0
展开全部 隐藏空白更改
内联 并排
Showing with 212 addition and 142 deletion
documentation20/en/02.getting-started/02.taosdemo/docs.md
浏览文件 @ 4311b0e5
Since TDengine was open sourced in July 2019, it has gained a lot of popularity among time-series database developers with its innovative data modeling design, simple installation method, easy programming interface, and powerful data insertion and query performance. The insertion and querying performance is often astonishing to users who are new to TDengine. In order to help users to experience the high performance and functions of TDengine in the shortest time, we developed an application called `taosBenchmark` (was named `taosdemo`) for insertion and querying performance testing of TDengine. Then user can easily simulate the scenario of a large number of devices generating a very large amount of data. User can easily manipulate the number of columns, data types, disorder ratio, and number of concurrent threads with taosBenchmark customized parameters.
Running taosBenchmark is very simple. Just download the [TDengine installation package](https://www.taosdata.com/cn/all-downloads/) or compiling the [TDengine code](https://github.com/taosdata/TDengine). It can be found and run in the installation directory or in the compiled results directory.
Running taosBenchmark is very simple. Just download the TDengine installation package (https://www.taosdata.com/cn/all-downloads/) or compiling the TDengine code yourself (https://github.com/taosdata/TDengine). It can be found and run in the installation directory or in the compiled results directory.
# To run an insertion test with taosBenchmark
To run an insertion test with taosBenchmark
--
Executing taosBenchmark without any parameters results in the following output.
```
$ taosBenchmark
......@@ -70,6 +70,7 @@ Query OK, 6 row(s) in set (0.002972s)
```
After pressing any key taosBenchmark will create the database test and super table meters and generate 10,000 sub-tables representing 10,000 individule meter devices that report data. That means they independently using the super table meters as a template according to TDengine data modeling best practices.
```
taos> use test;
Database changed.
......@@ -91,7 +92,9 @@ taos> show stables;
meters | 2021-08-27 11:21:01.209 | 4 | 2 | 10000 |
Query OK, 1 row(s) in set (0.001740s)
```
Then taosBenchmark generates 10,000 records for each meter device.
```
...
====thread[3] completed total inserted rows: 6250000, total affected rows: 6250000. 347626.22 records/second====
......@@ -108,9 +111,11 @@ Spent 18.0863 seconds to insert rows: 100000000, affected rows: 100000000 with 1
insert delay, avg: 28.64ms, max: 112.92ms, min: 9.35ms
```
The above information is the result of a real test on a normal PC server with 8 CPUs and 64G RAM. It shows that taosBenchmark inserted 100,000,000 (no need to count, 100 million) records in 18 seconds, or an average of 552,909,049 records per second.
TDengine also offers a parameter-bind interface for better performance, and using the parameter-bind interface (taosBenchmark -I stmt) on the same hardware for the same amount of data writes, the results are as follows.
```
...
......@@ -145,12 +150,13 @@ Spent 6.0257 seconds to insert rows: 100000000, affected rows: 100000000 with 16
insert delay, avg: 8.31ms, max: 860.12ms, min: 2.00ms
```
It shows that taosBenchmark inserted 100 million records in 6 seconds, with a much more higher insertion performance, 1,659,590 records wer inserted per second.
It shows that taosBenchmark inserted 100 million records in 6 seconds, with a much more higher insertion performance, 1,659,590 records wer inserted per second.
Because taosBenchmark is so easy to use, so we have extended it with more features to support more complex parameter settings for sample data preparation and validation for rapid prototyping.
The complete list of taosBenchmark command-line arguments can be displayed via taosBenchmark --help as follows.
```
$ taosBenchmark --help
......@@ -197,52 +203,70 @@ Report bugs to <support@taosdata.com>.
```
taosBenchmark's parameters are designed to meet the needs of data simulation. A few commonly used parameters are described below.
```
-I, --interface=INTERFACE The interface (taosc, rest, and stmt) taosBenchmark uses. Default is 'taosc'.
```
The performance difference between different interfaces of taosBenchmark has been mentioned earlier, the -I parameter is used to select different interfaces, currently taosc, stmt and rest are supported. The -I parameter is used to select different interfaces, currently taosc, stmt and rest are supported. taosc uses SQL statements to write data, stmt uses parameter binding interface to write data, and rest uses RESTful protocol to write data.
```
-T, --threads=NUMBER The number of threads. Default is 8.
```
The -T parameter sets how many threads taosBenchmark uses to synchronize data writes, so that multiple threads can squeeze as much processing power out of the hardware as possible.
```
-b, --data-type=DATATYPE The data_type of columns, default: FLOAT, INT, FLOAT.
-w, --binwidth=WIDTH The width of data_type 'BINARY' or 'NCHAR'. Default is 64
-l, --columns=COLUMNS The number of columns per record. Demo mode by default is 3 (float, int, float). Max values is 4095
```
As mentioned earlier, tadosdemo creates a typical meter data reporting scenario by default, with each device containing three columns. They are current, voltage and phases. TDengine supports BOOL, TINYINT, SMALLINT, INT, BIGINT, FLOAT, DOUBLE, BINARY, NCHAR, TIMESTAMP data types. By using -b with a list of types allows you to specify the column list with customized data type. Using -w to specify the width of the columns of the BINARY and NCHAR data types (default is 64). The -l parameter can be added to the columns of the data type specified by the -b parameter with the total number of columns of the INT type, which reduces the manual input process in case of a particularly large number of columns, up to 4095 columns.
```
-r, --rec-per-req=NUMBER The number of records per request. Default is 30000.
```
To reach TDengine performance limits, data insertion can be executed by using multiple clients, multiple threads, and batch data insertions at once. The -r parameter sets the number of records batch that can be stitched together in a single write request, the default is 30,000. The effective number of spliced records is also related to the client buffer size, which is currently 1M Bytes. If the record column width is large, the maximum number of spliced records can be calculated by dividing 1M by the column width (in bytes).
```
-t, --tables=NUMBER The number of tables. Default is 10000.
-n, --records=NUMBER The number of records per table. Default is 10000.
-M, --random The value of records generated are totally random. The default is to simulate power equipment scenario.
```
As mentioned earlier, taosBenchmark creates 10,000 tables by default, and each table writes 10,000 records. taosBenchmark can set the number of tables and the number of records in each table by -t and -n. The data generated by default without parameters are simulated real scenarios, and the simulated data are current and voltage phase values with certain jitter, which can more realistically show TDengine's efficient data compression ability. If you need to simulate the generation of completely random data, you can pass the -M parameter.
```
-y, --answer-yes Default input yes for prompt.
```
As we can see above, taosBenchmark outputs a list of parameters for the upcoming operation by default before creating a database or inserting data, so that the user can know what data is about to be written before inserting. To facilitate automatic testing, the -y parameter allows taosBenchmark to write data immediately after outputting the parameters.
```
-O, --disorder=NUMBER Insert order mode--0: In order, 1 ~ 50: disorder ratio. Default is in order.
-R, --disorder-range=NUMBER Out of order data's range, ms, default is 1000.
```
In some scenarios, the received data does not arrive in exact order, but contains a certain percentage of out-of-order data, which TDengine can also handle very well. In order to simulate the writing of out-of-order data, tadosdemo provides -O and -R parameters to be set. The -O parameter is the same as the -O parameter for fully ordered data writes. 1 to 50 is the percentage of data that contains out-of-order data. The -R parameter is the range of the timestamp offset of the out-of-order data, default is 1000 milliseconds. Also note that temporal data is uniquely identified by a timestamp, so garbled data may generate the exact same timestamp as previously written data, and such data may either be discarded (update 0) or overwrite existing data (update 1 or 2) depending on the update value created by the database, and the total number of data entries may not match the expected number of entries.
```
-g, --debug Print debug info.
```
If you are interested in the taosBenchmark insertion process or if the data insertion result is not as expected, you can use the -g parameter to make taosBenchmark print the debugging information in the process of the execution to the screen or import it to another file with the Linux redirect command to easily find the cause of the problem. In addition, taosBenchmark will also output the corresponding executed statements and debugging reasons to the screen after the execution fails. You can search the word "reason" to find the error reason information returned by the TDengine server.
```
-x, --aggr-func Test aggregation funtions after insertion.
```
TDengine is not only very powerful in insertion performance, but also in query performance due to its advanced database engine design. tadosdemo provides a -x function that performs the usual query operations and outputs the query consumption time after the insertion of data. The following is the result of a common query after inserting 100 million rows on the aforementioned server.
You can see that the select * fetch 100 million rows (not output to the screen) operation consumes only 1.26 seconds. The most of normal aggregation function for 100 million records usually takes only about 20 milliseconds, and even the longest count function takes less than 40 milliseconds.
```
taosBenchmark -I stmt -T 48 -y -x
...
......@@ -264,7 +288,9 @@ select min(current) took 0.025812 second(s)
select first(current) took 0.024105 second(s)
...
```
In addition to the command line approach, taosBenchmark also supports take a JSON file as an incoming parameter to provide a richer set of settings. A typical JSON file would look like this.
```
{
"filetype": "insert",
......@@ -273,17 +299,17 @@ In addition to the command line approach, taosBenchmark also supports take a JSO
"port": 6030,
"user": "root",
"password": "taosdata",
"thread_count": 4,
"thread_count_create_tbl": 4,
"result_file": "./insert_res.txt",
"confirm_parameter_prompt": "no",
"insert_interval": 0,
"interlace_rows": 100,
"thread_count": 4,
"thread_count_create_tbl": 4,
"result_file": "./insert_res.txt",
"confirm_parameter_prompt": "no",
"insert_interval": 0,
"interlace_rows": 100,
"num_of_records_per_req": 100,
"databases": [{
"dbinfo": {
"name": "db",
"drop": "yes",
"drop": "yes",
"replica": 1,
"days": 10,
"cache": 16,
......@@ -301,39 +327,41 @@ In addition to the command line approach, taosBenchmark also supports take a JSO
},
"super_tables": [{
"name": "stb",
"child_table_exists":"no",
"childtable_count": 100,
"childtable_prefix": "stb_",
"auto_create_table": "no",
"batch_create_tbl_num": 5,
"data_source": "rand",
"insert_mode": "taosc",
"insert_rows": 100000,
"childtable_limit": 10,
"childtable_offset":100,
"interlace_rows": 0,
"insert_interval":0,
"max_sql_len": 1024000,
"disorder_ratio": 0,
"disorder_range": 1000,
"timestamp_step": 10,
"start_timestamp": "2020-10-01 00:00:00.000",
"sample_format": "csv",
"sample_file": "./sample.csv",
"tags_file": "",
"child_table_exists":"no",
"childtable_count": 100,
"childtable_prefix": "stb_",
"auto_create_table": "no",
"batch_create_tbl_num": 5,
"data_source": "rand",
"insert_mode": "taosc",
"insert_rows": 100000,
"childtable_limit": 10,
"childtable_offset":100,
"interlace_rows": 0,
"insert_interval":0,
"max_sql_len": 1024000,
"disorder_ratio": 0,
"disorder_range": 1000,
"timestamp_step": 10,
"start_timestamp": "2020-10-01 00:00:00.000",
"sample_format": "csv",
"sample_file": "./sample.csv",
"tags_file": "",
"columns": [{"type": "INT"}, {"type": "DOUBLE", "count":10}, {"type": "BINARY", "len": 16, "count":3}, {"type": "BINARY", "len": 32, "count":6}],
"tags": [{"type": "TINYINT", "count":2}, {"type": "BINARY", "len": 16, "count":5}]
}]
}]
}
```
For example, we can specify different number of threads for table creation and data insertion with "thread_count" and "thread_count_create_tbl". You can use a combination of "child_table_exists", "childtable_limit" and "childtable_offset" to use multiple taosBenchmark processes (even on different computers) to write to different ranges of child tables of the same super table at the same time. You can also import existing data by specifying the data source as a csv file with "data_source" and "sample_file".
Use taosBenchmark for query and subscription testing
--
# Use taosBenchmark for query and subscription testing
taosBenchmark can not only write data, but also perform query and subscription functions. However, a taosBenchmark instance can only support one of these functions, not all three, and the configuration file is used to specify which function to test.
The following is the content of a typical query JSON example file.
```
{
"filetype": "query",
......@@ -373,7 +401,9 @@ The following is the content of a typical query JSON example file.
}
}
```
The following parameters are specific to the query in the JSON file.
```
"query_times": the number of queries per query type
"query_mode": query data interface, "tosc": call TDengine's c interface; "resetful": use restfule interface. Options are available. Default is "taosc".
......@@ -392,6 +422,7 @@ The following parameters are specific to the query in the JSON file.
```
The following is a typical subscription JSON example file content.
```
{
"filetype":"subscribe",
......@@ -404,34 +435,36 @@ The following is a typical subscription JSON example file content.
"confirm_parameter_prompt": "no",
"specified_table_query":
{
"concurrent":1,
"mode":"sync",
"interval":0,
"restart":"yes",
"concurrent":1,
"mode":"sync",
"interval":0,
"restart":"yes",
"keepProgress":"yes",
"sqls": [
{
"sql": "select * from stb00_0 ;",
"sql": "select * from stb00_0 ;",
"result": "./subscribe_res0.txt"
}]
},
"super_table_query":
"super_table_query":
{
"stblname": "stb0",
"threads":1,
"mode":"sync",
"interval":10000,
"restart":"yes",
"threads":1,
"mode":"sync",
"interval":10000,
"restart":"yes",
"keepProgress":"yes",
"sqls": [
{
"sql": "select * from xxxx where ts > '2021-02-25 11:35:00.000' ;",
"sql": "select * from xxxx where ts > '2021-02-25 11:35:00.000' ;",
"result": "./subscribe_res1.txt"
}]
}
}
```
The following are the meanings of the parameters specific to the subscription function.
```
"interval": interval for executing subscriptions, in seconds. Optional, default is 0.
"restart": subscription restart." yes": restart the subscription if it already exists, "no": continue the previous subscription. (Please note that the executing user needs to have read/write access to the dataDir directory)
......@@ -439,11 +472,12 @@ The following are the meanings of the parameters specific to the subscription fu
"resubAfterConsume": Used in conjunction with keepProgress to call unsubscribe after the subscription has been consumed the appropriate number of times and to subscribe again.
"result": the name of the file to which the query result is written. Optional, default is null, means the query result will not be written to the file. Note: The file to save the result after each sql statement cannot be renamed, and the file name will be appended with the thread number when generating the result file.
```
Conclusion
--
# Conclusion
TDengine is a big data platform designed and optimized for IoT, Telematics, Industrial Internet, DevOps, etc. TDengine shows a high performance that far exceeds similar products due to the innovative data storage and query engine design in the database kernel. And withSQL syntax support and connectors for multiple programming languages (currently Java, Python, Go, C#, NodeJS, Rust, etc. are supported), it is extremely easy to use and has zero learning cost. To facilitate the operation and maintenance needs, we also provide data migration and monitoring functions and other related ecological tools and software.
For users who are new to TDengine, we have developed rich features for taosBenchmark to facilitate technical evaluation and stress testing. This article is a brief introduction to taosBenchmark, which will continue to evolve and improve as new features are added to TDengine.
For users who are new to TDengine, we have developed rich features for taosBenchmark to facilitate technical evaluation and stress testing. This article is a brief introduction to taosBenchmark, which will continue to evolve and improve as new features are added to TDengine.
As part of TDengine, taosBenchmark's source code is fully open on the GitHub. Suggestions or advices about the use or implementation of taosBenchmark or TDengine are welcomed on GitHub or in the Taos Data user group.
documentation20/en/02.getting-started/docs.md
浏览文件 @ 4311b0e5
......@@ -22,7 +22,7 @@ TDengine is very easy to install, from download to successful installation in ju
<ul id="server-packageList" class="package-list"></ul>
For detailed installation steps, please refer to [How to install/uninstall TDengine with installation package](https://www.taosdata.com/en/getting-started/install).
For detailed installation steps, please refer to [How to install/uninstall TDengine with installation package](https://www.taosdata.com/getting-started/install).
**Click [here](https://github.com/taosdata/TDengine/releases) for release notes.**
......@@ -57,7 +57,7 @@ To run taosdump, you need to install the TDengine server or TDengine client inst
If you want to contribute to TDengine, please visit [TDengine GitHub page](https://github.com/taosdata/TDengine) for detailed instructions on build and installation from the source code.
**To download other components, beta, or early releases, please click [here](https://www.taosdata.com/cn/all-downloads/)**
**To download other components, beta version, or early releases, please click [here](https://www.taosdata.com/en/all-downloads/).**
## <a class="anchor" id="start"></a>Quick Launch
......
Markdown is supported
0% 或 .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑！
想要评论请 注册 或