Use custom log query and analysis - Investigation| Alibaba Cloud Documentation Center

Overview

In the Security Center console, choose Investigation > Log Analysis and enter SQL statements in Search & Analyze to perform custom log queries and analysis. A log query statement consists of two parts: search syntax and analytics syntax, divided by vertical bars (|). Syntax for query statements

You can perform custom log queries and analysis without the need to specify the search syntax or analytics syntax. The following list describes the search syntax and analytics syntax:

Search: You can use keywords, fuzzy match conditions, numeric values, ranges, or combinations to generate search conditions. If the value of the Search part is an asterisk (*) or is empty, data from the specified period is not filtered. In this case, all data from the specified period is used for analysis.
Analytics: It calculates and collects statistics on search results or full data. If the value of the Analytics part is empty, the results of the query are returned but no statistics are calculated.

Search syntax

The search syntax of Log Service supports both full-text query and field query. The query box supports features such as multi-line search and syntax highlighting.

Full-text query
You can enter keywords to search for logs without the need to specify fields. To use multiple keywords, enclose each keyword within a pair of quotation marks (") and separate them with spaces or the operator and. Examples:
- Query with multiple keywords specified
  Search for the logs that contain www.aliyun.com and 404. Examples:
```
www.aliyun.com 404
```
  or:
```
www.aliyun.com and 404
```
- Conditional query
  Search for the logs that contain www.aliyun.com and error, or the logs that contain www.aliyun.com and 404. Example:
```
www.aliyun.com and (error or 404)
```
- Suffix-based query
  Search for the logs that contain www.aliyun.com and start with failed_. Example:
```
www.aliyun.com and failed_*
```
  Note A full-text query supports only suffixes with asterisks (*). Prefixes with asterisks (*) are not supported.
Field query
Data of numeric fields can be filtered by using Field:Value or Field≥Value. You can combine the two formats by using and or or. It can also be used with a full-text query by using the combination specified by and and or.

Log Service supports more accurate queries based on fields.
- Multi-field query
  Search for alert logs whose severity level is serious. Example:
```
__topic__ : sas-security-log and level: serious
```
  Search for all SSH logons on the client whose IP address is 1.2.3.4. Example:
```
__topic__:aegis-log-login and ip:1.2.3.4 and warn_type:SSHLOGIN
```
  Note Each log contains the __topic__ field that indicates a log topic. Logs are distinguished by this field. In these examples, the fields, such as level, warn_type, and ip, are the fields for specific types of logs.
- Numeric field query
  Search for all internal DNS query logs that have a response time of more than 1 second. Example:
```
__topic__:local-dns and time_usecond > 1000000
```
  Range-based queries are also supported. For example, you can use range-based queries to search for internal DNS logs with a response time that is greater than 1 second but less than or equal to 10 seconds. Example:
```
__topic__:local-dns and time_usecond in [1000000,10000000]
```
For more information about the syntax, see Log search.

Analytics syntax

You can use SQL-92 statements to analyze and collect statistics on logs. For more information about the syntax and functions supported by Log Service, see Log analysis.

The FROM tablename clause in the standard SQL syntax can be omitted from analytics statements. This means that the FROM LOG clause can be omitted.

By default, Log Service returns the first 100 log entries. You can modify the number of log entries to return by using the LIMIT syntax. For more information, see LIMIT syntax.

Time-based log query and analysis

Each log entry has a built-in field __time__, which indicates the time at which this log is generated. This field facilitates time-based statistical analysis. The value of the field is a UNIX timestamp representing the number of seconds that have elapsed since the epoch time January 1, 1970, 00:00:00 UTC. Therefore, a timestamp must be converted into a supported format before it can be displayed.

Select and display the time
In this example, query the last 10 logon logs with the IP address of 1.2.3.4 within a specific time range. The return value includes the time, source IP address, and logon type. Example:
```
__topic__: aegis-log-login and ip: 1.2.3.4
| select date_format(__time__, '%Y-%m-%d %H:%i:%s') as time, warn_ip, warn_type 
order by __time__ desc 
limit 10
```
Calculate the time
Use __time__ to calculate the number of days after the logon. Example:
```
__topic__: aegis-log-login and ip: 1.2.3.4
| select date_format(__time__, '%Y-%m-%d %H:%i:%s') as time, warn_ip, warn_type ,
round((to_unixtime(now()) - __time__)/86400,1) as "days_passed"
order by __time__ desc 
limit 10
```
In this example, round((to_unixtime(now()) - __time__)/86400, 1) is used to perform the calculation. First, the function uses to_unixtime to convert the time returned by now() to a UNIX timestamp. Second, it subtracts the built-in __time__ field from the calculated value to obtain the number of seconds that have elapsed. Then, the function divides the calculated value by 86400, which is the total number of seconds in a day. Finally, the round(data, 1) function rounds the obtained value to one decimal place to calculate the number of days that have passed since the generation of each attack log.
Group statistics based on a specific time
If you want to know the logon trends for a device within a specific time range, execute the following SQL statement:
```
__topic__: aegis-log-login and ip: 1.2.3.4
| select date_trunc('day', __time__) as dt,
count(1) as PV
group by dt
order by dt
```
The built-in __time__ field is passed to the date_trunc('day', ..) function to align the time by day. Each log entry is grouped into the partition of the day to which it belongs to facilitate the calculation of the total number (count(1)). The log entries are sorted by partition time block. You can use other values for the first parameter of the date_trunc function to group log entries based on other time units, such as second, minute, hour, week, month, and year. For more information about the function, see Date and time functions.
Group statistics based on a flexible time
If you want to know more flexible rules of time grouping, such as logon trends per 5 minutes for devices of your account, execute the following SQL statement:
```
__topic__: aegis-log-login
| select from_unixtime(__time__ - __time__% 300) as dt,
count(1) as PV
group by dt
order by dt
limit 1000                        
```
In this example, the built-in time field is used to calculate __time__ - __time__% 300 and the from_unixtime function is used for formatting. Each log entry is grouped into a 5-minute (300 seconds) partition to facilitate the calculation of the total number (count(1)). The log entries are sorted by partition time block to obtain the first 1,000 log entries, which is equivalent to selecting data in the first 83 hours.

The following figure shows a line chart.

For more information about time-related functions, see Date and time functions. For example, the date_parse and date_format functions can convert a time format to another format.

Client IP address-based log query and analysis

The warn_ip field in a logon log entry indicates the source IP address of the logon.

Source country distribution of logons
Query the distribution of the source countries from which users log on to a server. Example:
```
__topic__: aegis-log-login and uuid: 12344567
| SELECT ip_to_country(warn_ip) as country,
count(1) as "Number of logons"
group by country
```
In this example, the ip_to_country function is used to retrieve the country that corresponds to warn_ip, which specifies the source IP address of the logon.

The following figure shows the distribution by using the world map.
Identity distribution of logons
Use the ip_to_province function to retrieve a more detailed distribution of logons based on provinces. Example:
```
__topic__: aegis-log-login and uuid: 12344567
| SELECT ip_to_province(warn_ip) as province,
         count(1) as "Number of logons"
         group by province
```
In this example, the ip_to_province function is used to retrieve the source province to which an IP address belongs. If the IP address is not from China, the system attempts to convert it to the province or state based on the country location of the IP address. However, if you select the China map, the province or state cannot be displayed.

The following figure shows the distribution by using the China map.
Geothermal distribution of logons
Use the ip_to_geo function to retrieve the geothermal distribution of logons:
```
__topic__: aegis-log-login and uuid: 12344567
| SELECT ip_to_geo(warn_ip) as geo,
         count(1) as "Number of logons"
         group by geo
         limit 10000
```
In this example, the ip_to_geo function is used to retrieve the latitude and longitude of an IP address. LIMIT is set to 10000 to retrieve the first 10,000 log entries.

The following figure shows the distribution by using AutoMap.

Note For more information about IP address-based features, see IP functions. For example, you can use the ip_to_provider function to obtain the provider of IP addresses and the ip_to_domain function to determine whether an IP address is public or private.