We’ve long recommended that customers use the Windows Firewall to protect SQL Server installations. Starting with Windows XP/SP2, and continuing with Windows Vista, the firewall has been enabled by default on Windows client operating systems. Windows Server 2008 marks the first time this protection has been extended to a Windows Server OS.
For those of you migrating from Windows Server 2003 or earlier to Windows Server 2008, if you have not previously heeded the advice to enable the firewall, you may be surprised by connectivity failures caused by the firewall (for any version of SQL) and you will need to take action to enable the connectivity you want.
Don’t panic! J Choosing the right firewall strategy isn’t as hard as it may seem, and it will pay dividends over the long run. We have a books online entry that has lots of good information on how to use the firewall. That document is available at
We strongly recommend you read that before making changes to your firewall strategy. For detailed information about the firewall, see
I use the term “firewall strategy” intentionally, because there are some tradeoffs that only you can make. You might want to simply “configure the firewall” to make it all “just work,” and you could do that, but it might expose you in ways you don’t intend. To highlight that point, I will refer you to a recent survey by David Litchfield of NGS Software. Quoting from the executive summary:
The survey found that there are approximately 368,000 Microsoft SQL Servers directly accessible on the Internet and around 124,000 Oracle database servers directly accessible on the Internet.
That’s a lot of servers directly exposed to the Internet, and I doubt strongly that level of exposure is intended. So we’re hopeful that you will spend some time making choices you are comfortable with, and that the change to enable the firewall on Windows Server 2008 will lead to a level of exposure that more closely matches your real business needs. And only you can determine your real business needs.
I should note that exposure by itself does not imply that there is any particular vulnerability. Indeed, only 4% of the exposed SQL servers were running a vulnerable version of SQL Server, and those few vulnerable servers appear to have gone unpatched for many years now (our competitors fared much worse in this regard, candidly). But in the event of a newly-discovered vulnerability in SQL Server those 368,000 servers could become 368,000 targets overnight (or faster), and we all want to avoid that.
So, our first piece of advice: review your existing firewall strategy, including your host and network firewalls, to ensure that none of your servers are unintentionally exposed to the internet or to untrustworthy insiders.
Step 0: Establish the scope of your scan, and appropriate policies. Will you be performing a simple audit of connectivity, or a full penetration test? What knowledge will the auditor have of your systems ahead of time? Who is authorized to have the results? Who is tasked with making changes based on the audit?
ISO-27002 discusses “ guidelines and general principles for initiating, implementing, maintaining, and improving information security management in an organization” and is worth reviewing if you don’t have established policies in place. Additional resources include
· The CERT OCTAVE Method
· The Open Source Security Testing Methodology
· The Center for Internet Security
· The Information Systems Audit and Control Association IS Auditing Procedure
Doing an audit carelessly can be costly and embarrassing. Review the resources above, and plan well.
Step 1: Locate your SQL Servers. Not only are SQL Servers in the datacenter serving enterprise-wide functions, but SQL Servers power desktop and department-wide applications outside the data center as well. And SQL Server comes in free editions like SQL Server Express, and SQL Server Express Advanced, which end users can download and install on their desktops, and which are bundled in other products like Visual Studio.
You can use SQL Browser to search for SQL Servers from different places on your network. Because Browser broadcasts to other instances of Browser in order to locate SQL Servers, there are some limitations.
· If the Browser service on a remote machine is not running, or is blocked by a firewall, this method won’t work of course
· If the firewall is enabled on the machine from which the scan is performed, that may interfere with gathering complete results. If a server takes more than 3 seconds to respond, that response may be blocked by the client-side firewall. See the UnicastResponsesToMulticastBroadcastDisabled property of the Windows firewall for more information.
· Because SQL Browser uses broadcasts, it can only “see” SQL servers in the same broadcast domain. So, you may have to do this from different places in your network.
· On clustered systems, be aware that browser is not a clustered resource, and does not support failover from one cluster node to the other.
For more information on Browser, see http://msdn.microsoft.com/en-us/library/ms181087.aspx.
Beyond using SQL Browser, you can perform a network scan looking for services listening on default ports for any of the SQL components. These default ports are
1434 for DB engine
4022, by convention
This is just the default ports for the most common scenarios, not an exhaustive list of possibilities. For more detail, see http://msdn.microsoft.com/en-us/library/cc646023(SQL.100).aspx.
Step 2: review the host firewall configuration to ensure it complies with your policy. For the Windows Firewall with Advanced Security, the basic checks are to ensure that there are appropriately configured exceptions for the services you intend to offer, and otherwise, ports in the firewall are blocked. In today’s highly networked world, with complex schemes to embed one protocol within another, and advanced scanning techniques designed to penetrate network firewalls, a host firewall is more important than ever.
Review the documents below for detailed information on using the firewall with various components of SQL:
Beyond just a simple “port open” or “port closed” decision, you should consider at least the network scope (all systems, local subnet, or a specific list of IP addresses), whether or not IP Sec is enabled, and what profile you are configuring. Review “Configuring the Windows Firewall to Allow SQL Server Access” for more information.
Step 3: Review your network firewall configuration to ensure that access to your network is correctly configured. Here, configuration choices vary widely and there are many makers of network firewall systems, including the Microsoft Internet Security and Acceleration (ISA) Server.
The ISA team has provided online training to help you design a perimeter firewall defense, covering the various classes of network firewalls, security, scalability, performance, and other topics. That material is available at
Step 4: Verify that your defenses are operating as expected. This may be the most difficult part because the job of the network is to deliver packets, and it may succeed despite your best efforts to stop it.
The basic idea is to attempt to establish connectivity to SQL Server from a location from which you believe the server is not accessible. Even if you have a machine that is providing a service to the whole world by design, it’s unlikely that you want to provide all services available on that machine to all comers. So make sure that you are checking for all the services the machine might offer.
It can help here to draw your network as a set of zones, and enumerate the connectivity policy within and between zones. A zone might be the “demilitarized zone” which traditionally provides a buffer between your public and private networks, or it may be “the sales zone,” where access is permitted to sales data. “The sales zone” does not consist of a list of authorized sales people, but rather a list of networks from which those sales people need access (after providing appropriate credentials of course).
As part of that process, you should identify zones from which a certain type of access is not permitted, and conduct a scan from that zone. This should fail, of course. It’s possible that all the networks you own are designed to permit connectivity to your database, and if that is the case, it means you will have to perform the scan from outside your network. A few things to keep in mind if you need to do that:
1. Don’t surprise the lawyers: be sure you have permission from the owners of both networks to perform such a scan.
2. Don’t surprise the network team: a scan can be resource intensive, and risky. Don’t do it without the approval of your network team.
3. Don’t surprise the security guys: be sure your own security team, and the security team of the network you are using, is aware of and prepared for the scan. When you set somebody’s pager off in the middle of the night because they think your network is under attack, we call that “an unfortunate event on your annual review.”
There are a variety of network scanners available to you to perform this scan. The Microsoft utility PortQry can help you do simple scans. The PortQry utility is available for download from the Microsoft Download Center. You may wish to consider third party tools like nmap, SuperScanner, Nesus, and Foundstone Enterprise for large scanning needs.
We have included some advice from our Application Consulting and Engineering team specifically about scanning:
· Use a moderate scanning speed when conducting UDP scans on the network, especially across WAN links; hosts that do not have a service on the port to which you are trying to connect will respond with “ICMP Port Unreachable”, and there have been instances where this can cause a denial of service on some routers
· Most scanning engines support tuning paramaters; try to break the network up logically and scan in smaller chunks to reduce possible outages and negative impact
· Use things like NetBIOS domain browse lists whenever possible to reduce the generation of traffic and “touching hosts”
If you’re conducting a complete network penetration test at the same time, things can get more complicated, but we’re not going to cover all of that here. Review the documents above, and plan for your pen test.
Step 5: Remedy the problems. No one technology may provide all the filtering capability you need. You may have to make tradeoffs and use a combination of network and host-based firewalls to achieve the protection you want.
-Shawn Hernan, SQL SIA Security, with much input from many people.
This posting is provided "AS IS" with no warranties, and confers no rights.
SDL alum, Shawn Hernan (now in the SQL Server team), has written an excellent post about SQL Server 2008,
Sounds like a movie title. No, it is actually a topic you need to know about if you plan to install SQL
Windows Server 2008 are și el firewall-ul activat by default, ca și Windows XP SP2 sau Windows Vista.
Did you know that Windows Server 2008 installation, by default, enables the Windows firewall to be ON
Ich hatte meine Erfahrungen bereits geschildert: Windows Server 2008 und der Default Zustand der Firewall Lese eben meine Feed und habe folgenden Artikel gefunden: SQL Server Security: SQL Server and the Windows Server 2008 Firewall via Michael Howard's
After I knew about the SQL 2008 RTM release news I have blogged SQL2008RTM-official here and as usual
Here are some of the links related to SQL Server 2005 on Windows Server 2008 Cluster: Support WebCast:
No matter what program, once installed on the PC, you often can find it out in the Add/Remove programs (we also called it currently installed programs list), and from where you often can uninstall the program you don't want.
after i moved from server 2k3 enterprise to sever 2008 enterprise and install SQL 2005 for my need, the remote users stopped connect to SQL, if I disable firewall in server2k8 every thing works fine but not with active firewall. I found way in this article (support.microsoft.com/.../de) by making a batch file, if the page is not longer active you can copy and paste bellow script in a notepad file and safe it as a batch file, i named it as OpenSQLPorts4Remotes-server2k8ONLY.bat (unusual long ;)). this is the script
netsh advfirewall firewall add rule name="Open Port 80" dir=in action=allow protocol=TCP localport=80
@echo ========= SQL Server Ports ===================
@echo Enabling SQLServer default instance port 1433
netsh advfirewall firewall add rule name="SQL Server" dir=in action=allow protocol=TCP localport=1433
@echo Enabling Dedicated Admin Connection port 1434
netsh advfirewall firewall add rule name="SQL Admin Connection" dir=in action=allow protocol=TCP localport=1434
@echo Enabling Conventional SQL Server Service Broker port 4022
netsh advfirewall firewall add rule name="SQL Service Broker" dir=in action=allow protocol=TCP localport=4022
@echo Enabling Transact SQL/RPC port 135
netsh advfirewall firewall add rule name="SQL Debugger/RPC" dir=in action=allow protocol=TCP localport=135
@echo ========= Analysis Services Ports ==============
@echo Enabling SSAS Default Instance port 2383
netsh advfirewall firewall add rule name="Analysis Services" dir=in action=allow protocol=TCP localport=2383
@echo Enabling SQL Server Browser Service port 2382
netsh advfirewall firewall add rule name="SQL Browser" dir=in action=allow protocol=TCP localport=2382
@echo ========= Misc Applications ==============
@echo Enabling HTTP port 80
netsh advfirewall firewall add rule name="HTTP" dir=in action=allow protocol=TCP localport=80
@echo Enabling SSL port 443
netsh advfirewall firewall add rule name="SSL" dir=in action=allow protocol=TCP localport=443
@echo Enabling port for SQL Server Browser Service's 'Browse' Button
netsh advfirewall firewall add rule name="SQL Browser" dir=in action=allow protocol=UDP localport=1434
@echo Allowing multicast broadcast response on UDP (Browser Service Enumerations OK)
netsh firewall set multicastbroadcastresponse ENABLE