Implementing “Row Level Security” (RLS) with “Filter Predicate” in SQL Server 20...

Implementing “Row Level Security” (RLS) with “Filter Predicate” in SQL Server 2016 – Part 1

For an Organization its Data is the most important thing, and the Org’s IT department takes lot of measures to make sure data does not fall in wrong hands. The DBA’s and Application programmers setup different layers of security on top of data so that the user is only able to see the Filtered data/rows from a table that he has access to. DB programmers normally create abstracted Views or Stored Procedures with complex logic on top of the Tables by using other master-mapping tables that contains the user-data access key. Sometimes these security logics are not upto the standards, and may have various versions on a database and thus it becomes difficult to track, update and make them fail-proof. So, with SQL Server 2016 the new Row Level Security feature is going to handle this Out of the Box and as a Standard.

RLS or Row Level Security is a feature that enables fine grained control over access to rows in a table, allowing you to easily control which users can access which data with complete transparency to the application.

With this feature rows are filtered based on the Execution Context of the Query, rather than the current user access rights. A secure logic can be created to determine which user can see which rows and restrict any kind of data (rows) by designing a flexible and robust Security policy for a table.

–> To setup Row Level Security (RLS) on a particular table it needs only these simple steps:

1. Create Users and Grant Read/SELECT access for a Particular table.

2. Create a new Inline Table-Valued Function that will contain the Filter Predicate for that table. This Function Predicate can be a sophisticated business logic with multiple JOINs or a simple WHERE ManagerCode = ‘userHR’.

3. Create a new Security Policy for this table and add the above Function (Filter) Predicate to it.

Please note: that these Functions & Security Policies should be unique for a table. So to create RLS for an another table, you will need to create separate Function & Security Policy.

–> Step 1.a. Let’s create some test accounts: I will create three users for:

1. The CEO, over-all admin of the company data.

2. HR department head

3. Finance department head

CREATE USER userCEO WITHOUT LOGIN;

GO

CREATE USER userHR WITHOUT LOGIN;

GO

CREATE USER userFin WITHOUT LOGIN;

GO

–> Create a sample table [dbo].[Employee]: with a self-referencing Manager ID column.

CREATE TABLE dbo.Employees (

[EmpCode] NVARCHAR(50), -- Employee ID

[EmpName] NVARCHAR(250), -- Employee/Manager Full Name

[Salary] MONEY, -- Fictious Salary

[MgrCode] NVARCHAR(50) -- Manager ID

);

GO

-> Now insert some test records:

-- Top Boss CEO

INSERT INTO dbo.Employees VALUES ('userCEO' , 'CEO Top Boss' , 800, NULL)

-- Next 2 levels under CEO

INSERT INTO dbo.Employees VALUES ('userHR' , 'HR User' , 700, 'userCEO');

INSERT INTO dbo.Employees VALUES ('userFin' , 'Finance User' , 600, 'userCEO');

-- Employees under Kevin

INSERT INTO dbo.Employees VALUES ('manojp' , 'Manoj Pandey' , 100, 'userHR');

INSERT INTO dbo.Employees VALUES ('saurabhs', 'Saurabh Sharma', 400, 'userHR');

INSERT INTO dbo.Employees VALUES ('deepakb' , 'Deepak Biswal' , 500, 'userHR');

-- Employees under Amy

INSERT INTO dbo.Employees VALUES ('keshavk' , 'Keshav K' , 200, 'userFin');

INSERT INTO dbo.Employees VALUES ('viveks' , 'Vivek S' , 300, 'userFin');

GO

–> Let’s check the records before applying “Row Level Security”:

SELECT * FROM dbo.Employees; -- 8 rows

GO

As a normal SEELCT and without RLS, it just ignores my Execution Context and execute the Query and return all the 8 rows.

–> The Traditional way to setup the Row Level Security till now was as follows (a simple example):

-- Stored Procedure with User-Name passed as parameter:

CREATE PROCEDURE dbo.uspGetEmployeeDetails (@userAccess NVARCHAR(50))

AS

BEGIN

SELECT *

FROM dbo.Employees

WHERE [MgrCode] = @userAccess

OR @userAccess = 'userCEO'; -- CEO, the admin should see all rows

END

GO

-- Execute the SP with different parameter values:

EXEC dbo.uspGetEmployeeDetails @userAccess = 'userHR' -- only 3 rows

GO

EXEC dbo.uspGetEmployeeDetails @userAccess = 'userFin' -- only 2 rows

GO

EXEC dbo.uspGetEmployeeDetails @userAccess = 'userCEO' -- all 8 rows

GO

The above method is prone to issues, like SQL Injection and any other user can apply other user’s User-Name and get the information that he is not allowed to see. With this type of method you have to apply another security layer at the application level so whenever a user executes the SP it gets executed with the same user’s User-Name.

–> The new Row Level Security feature let you:
– apply this security at the database level
– and there is no need to apply the WHERE clause filter for the User-Name.

This makes the security system more reliable and robust by reducing the surface area of your security system.

–> Step 1.b. Grant Read/SELECT access on the dbo.Employee table to all 3 users:

GRANT SELECT ON dbo.Employees TO userCEO;

GO

GRANT SELECT ON dbo.Employees TO userHR;

GO

GRANT SELECT ON dbo.Employees TO userFin;

GO

–> Step 2. Let’s create an Inline Table-Valued Function to write our Filter logic:

CREATE FUNCTION dbo.fn_SecurityPredicateEmployee(@mgrCode AS sysname)

RETURNS TABLE

WITH SCHEMABINDING

AS

RETURN SELECT 1 AS fn_SecurityPredicateEmployee_result

-- Predicate logic

WHERE @mgrCode = USER_NAME()

OR USER_NAME() = 'userCEO'; -- CEO, the admin should see all rows

GO

This function returns value 1 when:

– a row in the MgrCode (i.e. the Manager Code) column is the same as the user executing the query (@@mgrCode = USER_NAME())

– or if the user executing the query is the Top Boss user (USER_NAME() = ‘userCEO’)

–> Step 3. Create a security policy adding the function as a filter predicate:

CREATE SECURITY POLICY ManagerFilter

ADD FILTER PREDICATE dbo.fn_SecurityPredicateEmployee(MgrCode) -- Filter Column from dbo.Employee table

ON dbo.Employees

WITH (STATE = ON); -- The state must be set to ON to enable the policy.

GO

The above Security Policy takes the Filter Predicate Logic from the associated Function and applies it to the Query as a WHERE clause.

–> Now let’s again check the records after applying “Row Level Security”:

SELECT * FROM dbo.Employees; -- 0 rows,

GO

The simple “SELECT *” statement will fetch me zero rows after applying RLS, as my user ID is not configured to have access to any of those rows & Table.

–> And if you check in the Execution Plan of above SELECT statement without WHERE clause, it will show you the Filter Predicate that is added by the Security Policy defined in Step #3 for applying RLS on this table, which looks like this:

[TestManDB].[dbo].[Employee].[MgrCode]=user_name()
OR user_name()=N’userCEO’

–> Let’s check the 3 users we created and provided them customized access to the dbo.Employee table and rows in it:

-- Execute as our immediate boss userHR (3 rows):

EXECUTE AS USER = 'userHR';

SELECT * FROM dbo.Employees; -- 3 rows

REVERT;

GO

-- Execute as our immediate boss userFin:

EXECUTE AS USER = 'userFin';

SELECT * FROM dbo.Employees; -- 2 rows

REVERT;

GO

-- Execute as our Top boss userCEO (8):

EXECUTE AS USER = 'userCEO';

SELECT * FROM dbo.Employees; -- 8 rows

REVERT;

GO

–> The results of the above 3 SELECTs looks like this:

So, as you can see the three users we created resulted in different results:

– The HR & Finance users got just 3 & 2 rows, respectively.

– But the admin CEO user got all the 8 rows.

–> Thus, by using RLS Filter Predicate feature in SQL Server 2016 you can create your own customized Security by creating an Inline Table-Valued Function that is linked to the Security Policy for your Table.

–> Check Block Predicates with RLS in my [next post, Part 2].

–> Final Cleanup

DROP SECURITY POLICY [dbo].[ManagerFilter]

GO

DROP FUNCTION [dbo].[fn_SecurityPredicateEmployee]

GO

DROP TABLE [dbo].[Employee]

GO

DROP PROCEDURE dbo.uspGetEmployeeDetails

GO

Check the same demo on YouTube:

Check & Like my FB Page

Implementing “Row Level Security” (RLS) with “Filter Predicate” in SQL Server 2016 – Part 1

Recommend

快递诈骗手法又升级！有人被骗近万元，收快递要小心7个套路

DDR5-4800 trades blows with DDR4-3200 in leaked Alder Lake benchmarks | TechSpot

高效能的个人SWOT分析

Voltage manipulation can bypass hardware security on AMD's server CPUs | TechSpo...

王团长区块链日记1376篇：回调了

Astro and US Army drop Call of Duty League sponsorship | TechSpot

Share Pie: 隐藏的DDD宝藏 -Nick

中国移动：5G建设高峰至2022年，全面推进回A股

The Janitor from Columbia · Kaushik Gopal's Site

六边形架构教程：构建可维护的Web应用程序 - DEV

About Joyk