The answer to your questions about policies is that it requires a joint
effort to establish the policies. You need the input of the developers as
to what it the application should be doing during its normal operation. The
system administrators need to accommodate the developers with a degree of
accessibility to allow the application to function, but never with total
access to the system (absolute access is never a good idea). The network
administrators configure the network access according to the agreed upon
access requirements. However, this process is usually agreed upon during
the actual discussion about the application before the code is even written
and not at deployment time.
After an application is deployed the admins should be doing what their jobs
require them to do: check the logs, patch systems, etc. There's no reason
why their roles or responsibilities should need to be changed for a single
application.
As for security: firewalls come from everything from the garden variety dumb
as bricks models that only feature ip/port restrictions to the more advanced
models that use technologies to inspect the traffic before it hits the
application and make sure that what is being transmitted falls within
parameters that would be considered "normal". The latter models have to
understand the protocol being used such as HTTP, FTP, SOAP, etc. SSL can be
decrypted by an SSL accelerator applicance making the encryption not a
problem for firewall appliances. Application level proxies also can help
with security, but require more configuration and are more obvious than the
cut-through variety that use SMLI or SPF.
Preventing a total system takeover from a poorly written application
requires the usual amount of common sense: the principal of least security
(never give anything more privileges than it needs). Different user
accounts used for different applications provides a level of segregation
between the applications be it for file system or database access.
I'm not really sure if there's really a one stop source on what you're
looking for because your questions span three different security categories:
network security (covered in network administration books), system security
(covered in systems administration books), and designing secure applications
(covered in programming books). Then, if there's a database involved (as is
usually the case), you have a dba whose job revolves around tweaking the
performance and the security of the rdbms (oracle, sql server, db2, etc).
The reason for this segregation is usually because the roles in an
administration are segregated and that's not necessarilly a bad thing.
Generally, the admins are concerned with securing their appliances and
systems to the highest state possible while the dba and the programmer are
concerned with their respective applications. It's a lot to juggle.
How much lattitude or how many roles anyone has (particularly the developer)
is more a matter of the policy that you're working for making it completely
arbitrary. There are better practices and guidance available in each field,
but knowing everything takes hours and hours of reading plus years of
experience. If you're looking at trying your hand in seeing how in depth
you can go with programming and security grab some network gear and cheap
systems off EBay, construct a lab of your own, and go to town because
there's no substitute for doing.
On Wed, Mar 26, 2008 at 10:35 AM, william fitzgerald <wfitzgerald (at) tssg (dot) org [email concealed]>
wrote:
> Thanks Geoffrey,
>
> I agree firewalls are an important point of demarcation narrowing access
> (intended access) between various intranets.
>
> In your experience what kinds of rules or policies are put in place when
> an enterprise application goes live? What are the kinds of policies that
> application developers demand from system administrators?
>
> My argument is that it appears to me that (secure) Enterprise Web
> Service applications, particularly those involving access control, are
> typically focused at the application-domain only, rather than taking a
> more holistic approach to also include the underlying infrastructure
> (for example, firewalls). As a result, infrastructure configurations may
> unintentionally hinder and prohibit the normal operation of the Web
> Service.
>
> Thus, the ideal firewall configuration is one that is aligned with the
> application supported by the system, that is, it permits valid
> application traffic, and, preferably, no more and no less.
>
> As I stated in my original post, Web Service developers assume the
> underlying infrastructure is automatically available. Also there seems
> to be a tendency to tunnel (for example SOAP) over http or https. From
> this point of view, Web Service developers may form the opinion that
> firewalls are redundant as they typically have ports 80 and 443
> accessible (and forward traffic to specialized user-space programs for
> further packet processing).
>
> Maybe this is correct! comments?
>
> In my opinion, deploying a network level firewall (such as Linux
> Netfilter) provisioned for Enterprise Web Services is not simply about
> opening port 80 on the server for all traffic; one may wish to deny
> certain nodes (IP addresses, etc.), only accept HTTP traffic from some
> nodes, require other nodes to use HTTPS and also deal with HTTP traffic
> that is tunneled through proxies available on other ports.
>
> Comments?
>
> While low level infrastructure such as network firewalls may not solve
> all security issues ( as a more suitable application based XML firewall
> mite) in regard to Web Service applications, I believe as stated in my
> previous post, they have a role to play in applying the belt-and-braces
> approach to security best practices.
>
> Comments?
>
> What I am really looking for is some concrete documents, publications,
> administrator experience that helps clarify the important role of
> Network Access Controls (firewalls, IPS etc) within an enterprise SOA
> environment, if any.
>
> regards,
> Will.
>
> Geoffrey Gowey wrote:
> > To be succinct: yes, firewalls are still very much a necessity. If one
> > ignores the more advanced features of firewalls and only focuses on the
> > basic features of ip packet source/destination filtering there's still
> > evidence of a clear need for them before even seeing what the usage of
> > other technologies such as SMLI or SPF bring to the table. Relying on
> > operating system and application security alone is insufficient in a
> > networked environment.
> >
> > By putting internal and external firewalls in place, a corporation has
> > choke points that can enforce corporate policy of what an application
> > really should be limited to doing. A for instance would be the
> > limitation of IP access from a DMZ web server to a middle tier server
> > for its data. A firewall between the DMZ server and the middle tier
> > server can be configured to ensure that this is the only access the
> > server has and not access to other systems that it has no reason to be
> > interacting with (such as a mail server). This helps reduce exposure
> > should the DMZ server be compromised. Also, if the firewall between the
> > DMZ server and the middle tier server is configured properly, you will
> > be able to discover irregular activity that could indicate a misbehaving
> > application or a server compromise by traffic trying to conduct actions
> > that were not part of the agreed on policy.
> >
> >
> > On Tue, Mar 25, 2008 at 4:56 AM, william fitzgerald
> > <wfitzgerald (at) tssg (dot) org [email concealed] <mailto:wfitzgerald (at) tssg (dot) org [email concealed]>> wrote:
> >
> > Dear Firewall Experts,
> >
> > Provocative Question:
> > ++++++++++++++++++++
> > Are firewalls obsolete in a world involving enterprise Web Service
> SOA?
> >
> > What do I mean by the above question: given that Web Services (J2EE
> and
> > so forth) tend to tunnel through http and https (eg. SOAP) what role
> can
> > a traditional network firewall play? If its just a matter of opening
> > ports http and https for your dedicated enterprise services then is
> > there even a need for a firewall!
> >
> > I am asking this question not to be flamed but to provoke a
> discussion
> > as to why we still need firewalls.
> >
> > Assumptions:
> > ++++++++++++
> > I use the term firewall loosely to mean "network access control".
> That
> > is, its a mechanism to prevent unwanted packets. Therefore, a
> firewall
> > could be iptables (stateful, DPI etc) or even the proxy TCP
> Wrappers,
> > cisco and so forth.
> >
> > In particular, I have focused on Linux iptables and TCP Wrapper. I
> > realize that one can install an xml based firewall to inspect packet
> > content in regard to web services.
> >
> > Scenario Network:
> > ++++++++++++++++++
> > Internet ---> Firewall ---> Enterprise SOA Server ---> Additional
> > firewalls and back-end database servers etc.
> >
> > Is it a case that in this Enterprise SOA environment the NAC
> firewall is
> > made redundant (as opposed to an xml firewall):
> >
> > Internet ---> Enterprise SOA Webservice server
> >
> > Assuming of course the servers are dedicated Web Service servers
> that
> > run no other services such as DHCP, intranet web server, email and
> so
> > forth that need to be protected?
> >
> > Firewall Justification:
> > +++++++++++++++++++++++
> >
> > I am trying to find publications, white papers, reports etc that
> state
> > the case for the need for firewalls. I need something concrete.
> >
> > The current information I have found (web service orientated!) tends
> to
> > say firewalls are obsolete when talking about enterprise SOA given
> that
> > once port 80 and 443 is open on the firewall the SOS services are
> > exposed and hence protection happens at the application layer of the
> > particular service.
> >
> > However, best practice suggests one should take a more holistic
> approach
> > to security and apply the "belt-and-braces" approach. That is,
> install
> > firewalls, IDS, AV, proper authentication at various OSI stack
> layers
> > etc etc. So we get a layered security affect, thus there must be a
> > justification for using a firewall still.
> >
> > My Opinion:
> > +++++++++++
> >
> > My opinion on what NAC firewalls can offer to web service SOA other
> than
> > simply opening port http and https is as follows:
> >
> > 1) control access to those ports via ip address ranges (eg.
> > customer/business subscribers)
> > 2) deep packet inspection to solicit appropriate content incoming
> and
> > outgoing from the SOA enterprise servers.
> > 3) ???? what else would be done? please comment.
> >
> > While I agree that there are xml based firewalls to monitor xml
> based
> > Web Service traffic, I wonder can it perform access controls at the
> > lower levels like network based firewalls (for example, block
> certain IP
> > addresses)? My guess is they don't given the operate at the
> application
> > layer.
> >
> > I also wonder why one would invest in an xml firewall that is
> dedicated
> > to one kind of traffic profiling and not use for example a very
> > expensive cisco firewall that can cover a multitude of traffic
> > profiling. Presumably these expensive firewalls (or the equivalent
> > unexpensive iptables firewall) can inspect the packet for malicious
> > content to and from the enterprise servers (I believe we have
> > snort-2-iptables to also help here). At any rate, I do not want to
> start
> > a huge debate on the pros and cons of an xml firewall versus a
> network
> > firewall as I am aware dedicated firewalls specialize in various
> traffic
> > profiling. Also its best practice to install a wide range for
> firewall
> > capabilities.
> >
> > The real issue is the justification of NAC's in an enterprise SOA
> > environment. Of course, if this enterprise environment also included
> the
> > company standard services such as email, dns, web server etc I can
> see
> > the major impact of the NAC firewall. But what is the case for
> dedicated
> > enterprise SOA?
> >
> >
> > My shortcomings:
> > ++++++++++++++++
> > My inexperience in an enterprise network environment of how things
> are
> > really carried out rather than what is done in theory.
> >
> >
> > Summary:
> > ++++++++
> > What role do NAC's have to play in an environment of enterprise
> > application services?
> >
> > All pointers to documentation and your comments are welcome.
> >
> > I look forward to your support,
> > regards,
> > Will.
> >
> > --
> > William M. Fitzgerald,
> > PhD Student,
> > Telecommunications Software & Systems Group,
> > ArcLabs Research and Innovation Centre,
> > Waterford Institute of Technology,
> > WIT West Campus,
> > Carriganore,
> > Waterford.
> > Office Ph: +353 51 302937
> > Mobile Ph: +353 87 9527083
> > Web: www.williamfitzgerald.org <http://www.williamfitzgerald.org>
> > www.linkedin.com/in/williamfitzgerald
> > <http://www.linkedin.com/in/williamfitzgerald>
> > www.ryze.com/go/wfitzgerald <http://www.ryze.com/go/wfitzgerald
> >
> >
> >
> >
> >
> >
> >
> > --
> > Kindest Regards,
> >
> > Geoff
>
> --
> William M. Fitzgerald,
> PhD Student,
> Telecommunications Software & Systems Group,
> ArcLabs Research and Innovation Centre,
> Waterford Institute of Technology,
> WIT West Campus,
> Carriganore,
> Waterford.
> Office Ph: +353 51 302937
> Mobile Ph: +353 87 9527083
> Web: www.williamfitzgerald.org
> www.linkedin.com/in/williamfitzgerald
> www.ryze.com/go/wfitzgerald
>
>
>
>
--
Kindest Regards,
Geoff
The answer to your questions about policies is that it requires a joint effort to
establish the policies. You need the input of the developers as to
what it the application should be doing during its normal operation.
The system administrators need to accommodate the developers with a
degree of accessibility to allow the application to function, but never with total access to the system (absolute access is never a good idea). The network administrators configure the network access according to the agreed upon access requirements. However, this process is usually agreed upon during the actual discussion about the application before the code is even written and not at deployment time.<br>
<br>After an application is deployed the admins should be doing what their jobs require them to do: check the logs, patch systems, etc. There's no reason why their roles or responsibilities should need to be changed for a single application. <br>
<br>As for security: firewalls come from everything from the garden variety dumb as bricks models that only feature ip/port restrictions to the more advanced models that use technologies to inspect the traffic before it hits the application and make sure that what is being transmitted falls within parameters that would be considered "normal". The latter models have to understand the protocol being used such as HTTP, FTP, SOAP, etc. SSL can be decrypted by an SSL accelerator applicance making the encryption not a problem for firewall appliances. Application level proxies also can help with security, but require more configuration and are more obvious than the cut-through variety that use SMLI or SPF. <br>
<br>Preventing a total system takeover from a poorly written application requires the usual amount of common sense: the principal of least security (never give anything more privileges than it needs). Different user accounts used for different applications provides a level of segregation between the applications be it for file system or database access.<br>
<br>I'm not really sure if there's really a one stop source on what you're looking for because your questions span three different security categories: network security (covered in network administration books), system security (covered in systems administration books), and designing secure applications (covered in programming books). Then, if there's a database involved (as is usually the case), you have a dba whose job revolves around tweaking the performance and the security of the rdbms (oracle, sql server, db2, etc). The reason for this segregation is usually because the roles in an administration are segregated and that's not necessarilly a bad thing. Generally, the admins are concerned with securing their appliances and systems to the highest state possible while the dba and the programmer are concerned with their respective applications. It's a lot to juggle.<br>
<br>How much lattitude or how many roles anyone has (particularly the developer) is more a matter of the policy that you're working for making it completely arbitrary. There are better practices and guidance available in each field, but knowing everything takes hours and hours of reading plus years of experience. If you're looking at trying your hand in seeing how in depth you can go with programming and security grab some network gear and cheap systems off EBay, construct a lab of your own, and go to town because there's no substitute for doing.<br>
<br><br><div class="gmail_quote">On Wed, Mar 26, 2008 at 10:35 AM, william fitzgerald <<a href="mailto:wfitzgerald (at) tssg (dot) org [email concealed]">wfitzgerald (at) tssg (dot) org [email concealed]</a>> wrote:<br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Thanks Geoffrey,<br>
<br>
I agree firewalls are an important point of demarcation narrowing access<br>
(intended access) between various intranets.<br>
<br>
In your experience what kinds of rules or policies are put in place when<br>
an enterprise application goes live? What are the kinds of policies that<br>
application developers demand from system administrators?<br>
<br>
My argument is that it appears to me that (secure) Enterprise Web<br>
Service applications, particularly those involving access control, are<br>
typically focused at the application-domain only, rather than taking a<br>
more holistic approach to also include the underlying infrastructure<br>
(for example, firewalls). As a result, infrastructure configurations may<br>
unintentionally hinder and prohibit the normal operation of the Web Service.<br>
<br>
Thus, the ideal firewall configuration is one that is aligned with the<br>
application supported by the system, that is, it permits valid<br>
application traffic, and, preferably, no more and no less.<br>
<br>
As I stated in my original post, Web Service developers assume the<br>
underlying infrastructure is automatically available. Also there seems<br>
to be a tendency to tunnel (for example SOAP) over http or https. From<br>
this point of view, Web Service developers may form the opinion that<br>
firewalls are redundant as they typically have ports 80 and 443<br>
accessible (and forward traffic to specialized user-space programs for<br>
further packet processing).<br>
<br>
Maybe this is correct! comments?<br>
<br>
In my opinion, deploying a network level firewall (such as Linux<br>
Netfilter) provisioned for Enterprise Web Services is not simply about<br>
opening port 80 on the server for all traffic; one may wish to deny<br>
certain nodes (IP addresses, etc.), only accept HTTP traffic from some<br>
nodes, require other nodes to use HTTPS and also deal with HTTP traffic<br>
that is tunneled through proxies available on other ports.<br>
<br>
Comments?<br>
<br>
While low level infrastructure such as network firewalls may not solve<br>
all security issues ( as a more suitable application based XML firewall<br>
mite) in regard to Web Service applications, I believe as stated in my<br>
previous post, they have a role to play in applying the belt-and-braces<br>
approach to security best practices.<br>
<br>
Comments?<br>
<br>
What I am really looking for is some concrete documents, publications,<br>
administrator experience that helps clarify the important role of<br>
Network Access Controls (firewalls, IPS etc) within an enterprise SOA<br>
environment, if any.<br>
<br>
regards,<br>
Will.<br>
<div class="Ih2E3d"><br>
Geoffrey Gowey wrote:<br>
> To be succinct: yes, firewalls are still very much a necessity. If one<br>
> ignores the more advanced features of firewalls and only focuses on the<br>
> basic features of ip packet source/destination filtering there's still<br>
> evidence of a clear need for them before even seeing what the usage of<br>
> other technologies such as SMLI or SPF bring to the table. Relying on<br>
> operating system and application security alone is insufficient in a<br>
> networked environment.<br>
><br>
> By putting internal and external firewalls in place, a corporation has<br>
> choke points that can enforce corporate policy of what an application<br>
> really should be limited to doing. A for instance would be the<br>
> limitation of IP access from a DMZ web server to a middle tier server<br>
> for its data. A firewall between the DMZ server and the middle tier<br>
> server can be configured to ensure that this is the only access the<br>
> server has and not access to other systems that it has no reason to be<br>
> interacting with (such as a mail server). This helps reduce exposure<br>
> should the DMZ server be compromised. Also, if the firewall between the<br>
> DMZ server and the middle tier server is configured properly, you will<br>
> be able to discover irregular activity that could indicate a misbehaving<br>
> application or a server compromise by traffic trying to conduct actions<br>
> that were not part of the agreed on policy.<br>
><br>
><br>
> On Tue, Mar 25, 2008 at 4:56 AM, william fitzgerald<br>
</div><div><div></div><div class="Wj3C7c">> <<a href="mailto:wfitzgerald (at) tssg (dot) org [email concealed]">wfitzgerald (at) tssg (dot) org [email concealed]</a> <mailto:<a href="mailto:wfitzgerald (at) tssg (dot) org [email concealed]">wfitzgerald (at) tssg (dot) org [email concealed]</a>>> wrote:<br>
><br>
> Dear Firewall Experts,<br>
><br>
> Provocative Question:<br>
> ++++++++++++++++++++<br>
> Are firewalls obsolete in a world involving enterprise Web Service SOA?<br>
><br>
> What do I mean by the above question: given that Web Services (J2EE and<br>
> so forth) tend to tunnel through http and https (eg. SOAP) what role can<br>
> a traditional network firewall play? If its just a matter of opening<br>
> ports http and https for your dedicated enterprise services then is<br>
> there even a need for a firewall!<br>
><br>
> I am asking this question not to be flamed but to provoke a discussion<br>
> as to why we still need firewalls.<br>
><br>
> Assumptions:<br>
> ++++++++++++<br>
> I use the term firewall loosely to mean "network access control". That<br>
> is, its a mechanism to prevent unwanted packets. Therefore, a firewall<br>
> could be iptables (stateful, DPI etc) or even the proxy TCP Wrappers,<br>
> cisco and so forth.<br>
><br>
> In particular, I have focused on Linux iptables and TCP Wrapper. I<br>
> realize that one can install an xml based firewall to inspect packet<br>
> content in regard to web services.<br>
><br>
> Scenario Network:<br>
> ++++++++++++++++++<br>
> Internet ---> Firewall ---> Enterprise SOA Server ---> Additional<br>
> firewalls and back-end database servers etc.<br>
><br>
> Is it a case that in this Enterprise SOA environment the NAC firewall is<br>
> made redundant (as opposed to an xml firewall):<br>
><br>
> Internet ---> Enterprise SOA Webservice server<br>
><br>
> Assuming of course the servers are dedicated Web Service servers that<br>
> run no other services such as DHCP, intranet web server, email and so<br>
> forth that need to be protected?<br>
><br>
> Firewall Justification:<br>
> +++++++++++++++++++++++<br>
><br>
> I am trying to find publications, white papers, reports etc that state<br>
> the case for the need for firewalls. I need something concrete.<br>
><br>
> The current information I have found (web service orientated!) tends to<br>
> say firewalls are obsolete when talking about enterprise SOA given that<br>
> once port 80 and 443 is open on the firewall the SOS services are<br>
> exposed and hence protection happens at the application layer of the<br>
> particular service.<br>
><br>
> However, best practice suggests one should take a more holistic approach<br>
> to security and apply the "belt-and-braces" approach. That is, install<br>
> firewalls, IDS, AV, proper authentication at various OSI stack layers<br>
> etc etc. So we get a layered security affect, thus there must be a<br>
> justification for using a firewall still.<br>
><br>
> My Opinion:<br>
> +++++++++++<br>
><br>
> My opinion on what NAC firewalls can offer to web service SOA other than<br>
> simply opening port http and https is as follows:<br>
><br>
> 1) control access to those ports via ip address ranges (eg.<br>
> customer/business subscribers)<br>
> 2) deep packet inspection to solicit appropriate content incoming and<br>
> outgoing from the SOA enterprise servers.<br>
> 3) ???? what else would be done? please comment.<br>
><br>
> While I agree that there are xml based firewalls to monitor xml based<br>
> Web Service traffic, I wonder can it perform access controls at the<br>
> lower levels like network based firewalls (for example, block certain IP<br>
> addresses)? My guess is they don't given the operate at the application<br>
> layer.<br>
><br>
> I also wonder why one would invest in an xml firewall that is dedicated<br>
> to one kind of traffic profiling and not use for example a very<br>
> expensive cisco firewall that can cover a multitude of traffic<br>
> profiling. Presumably these expensive firewalls (or the equivalent<br>
> unexpensive iptables firewall) can inspect the packet for malicious<br>
> content to and from the enterprise servers (I believe we have<br>
> snort-2-iptables to also help here). At any rate, I do not want to start<br>
> a huge debate on the pros and cons of an xml firewall versus a network<br>
> firewall as I am aware dedicated firewalls specialize in various traffic<br>
> profiling. Also its best practice to install a wide range for firewall<br>
> capabilities.<br>
><br>
> The real issue is the justification of NAC's in an enterprise SOA<br>
> environment. Of course, if this enterprise environment also included the<br>
> company standard services such as email, dns, web server etc I can see<br>
> the major impact of the NAC firewall. But what is the case for dedicated<br>
> enterprise SOA?<br>
><br>
><br>
> My shortcomings:<br>
> ++++++++++++++++<br>
> My inexperience in an enterprise network environment of how things are<br>
> really carried out rather than what is done in theory.<br>
><br>
><br>
> Summary:<br>
> ++++++++<br>
> What role do NAC's have to play in an environment of enterprise<br>
> application services?<br>
><br>
> All pointers to documentation and your comments are welcome.<br>
><br>
> I look forward to your support,<br>
> regards,<br>
> Will.<br>
><br>
> --<br>
> William M. Fitzgerald,<br>
> PhD Student,<br>
> Telecommunications Software & Systems Group,<br>
> ArcLabs Research and Innovation Centre,<br>
> Waterford Institute of Technology,<br>
> WIT West Campus,<br>
> Carriganore,<br>
> Waterford.<br>
> Office Ph: +353 51 302937<br>
> Mobile Ph: +353 87 9527083<br>
</div></div>> Web: <a href="http://www.williamfitzgerald.org" target="_blank">www.williamfitzgerald.org</a> <<a href="http://www.williamfitzgerald.org" target="_blank">http://www.williamfitzgerald.org</a>><br>
effort to establish the policies. You need the input of the developers as
to what it the application should be doing during its normal operation. The
system administrators need to accommodate the developers with a degree of
accessibility to allow the application to function, but never with total
access to the system (absolute access is never a good idea). The network
administrators configure the network access according to the agreed upon
access requirements. However, this process is usually agreed upon during
the actual discussion about the application before the code is even written
and not at deployment time.
After an application is deployed the admins should be doing what their jobs
require them to do: check the logs, patch systems, etc. There's no reason
why their roles or responsibilities should need to be changed for a single
application.
As for security: firewalls come from everything from the garden variety dumb
as bricks models that only feature ip/port restrictions to the more advanced
models that use technologies to inspect the traffic before it hits the
application and make sure that what is being transmitted falls within
parameters that would be considered "normal". The latter models have to
understand the protocol being used such as HTTP, FTP, SOAP, etc. SSL can be
decrypted by an SSL accelerator applicance making the encryption not a
problem for firewall appliances. Application level proxies also can help
with security, but require more configuration and are more obvious than the
cut-through variety that use SMLI or SPF.
Preventing a total system takeover from a poorly written application
requires the usual amount of common sense: the principal of least security
(never give anything more privileges than it needs). Different user
accounts used for different applications provides a level of segregation
between the applications be it for file system or database access.
I'm not really sure if there's really a one stop source on what you're
looking for because your questions span three different security categories:
network security (covered in network administration books), system security
(covered in systems administration books), and designing secure applications
(covered in programming books). Then, if there's a database involved (as is
usually the case), you have a dba whose job revolves around tweaking the
performance and the security of the rdbms (oracle, sql server, db2, etc).
The reason for this segregation is usually because the roles in an
administration are segregated and that's not necessarilly a bad thing.
Generally, the admins are concerned with securing their appliances and
systems to the highest state possible while the dba and the programmer are
concerned with their respective applications. It's a lot to juggle.
How much lattitude or how many roles anyone has (particularly the developer)
is more a matter of the policy that you're working for making it completely
arbitrary. There are better practices and guidance available in each field,
but knowing everything takes hours and hours of reading plus years of
experience. If you're looking at trying your hand in seeing how in depth
you can go with programming and security grab some network gear and cheap
systems off EBay, construct a lab of your own, and go to town because
there's no substitute for doing.
On Wed, Mar 26, 2008 at 10:35 AM, william fitzgerald <wfitzgerald (at) tssg (dot) org [email concealed]>
wrote:
> Thanks Geoffrey,
>
> I agree firewalls are an important point of demarcation narrowing access
> (intended access) between various intranets.
>
> In your experience what kinds of rules or policies are put in place when
> an enterprise application goes live? What are the kinds of policies that
> application developers demand from system administrators?
>
> My argument is that it appears to me that (secure) Enterprise Web
> Service applications, particularly those involving access control, are
> typically focused at the application-domain only, rather than taking a
> more holistic approach to also include the underlying infrastructure
> (for example, firewalls). As a result, infrastructure configurations may
> unintentionally hinder and prohibit the normal operation of the Web
> Service.
>
> Thus, the ideal firewall configuration is one that is aligned with the
> application supported by the system, that is, it permits valid
> application traffic, and, preferably, no more and no less.
>
> As I stated in my original post, Web Service developers assume the
> underlying infrastructure is automatically available. Also there seems
> to be a tendency to tunnel (for example SOAP) over http or https. From
> this point of view, Web Service developers may form the opinion that
> firewalls are redundant as they typically have ports 80 and 443
> accessible (and forward traffic to specialized user-space programs for
> further packet processing).
>
> Maybe this is correct! comments?
>
> In my opinion, deploying a network level firewall (such as Linux
> Netfilter) provisioned for Enterprise Web Services is not simply about
> opening port 80 on the server for all traffic; one may wish to deny
> certain nodes (IP addresses, etc.), only accept HTTP traffic from some
> nodes, require other nodes to use HTTPS and also deal with HTTP traffic
> that is tunneled through proxies available on other ports.
>
> Comments?
>
> While low level infrastructure such as network firewalls may not solve
> all security issues ( as a more suitable application based XML firewall
> mite) in regard to Web Service applications, I believe as stated in my
> previous post, they have a role to play in applying the belt-and-braces
> approach to security best practices.
>
> Comments?
>
> What I am really looking for is some concrete documents, publications,
> administrator experience that helps clarify the important role of
> Network Access Controls (firewalls, IPS etc) within an enterprise SOA
> environment, if any.
>
> regards,
> Will.
>
> Geoffrey Gowey wrote:
> > To be succinct: yes, firewalls are still very much a necessity. If one
> > ignores the more advanced features of firewalls and only focuses on the
> > basic features of ip packet source/destination filtering there's still
> > evidence of a clear need for them before even seeing what the usage of
> > other technologies such as SMLI or SPF bring to the table. Relying on
> > operating system and application security alone is insufficient in a
> > networked environment.
> >
> > By putting internal and external firewalls in place, a corporation has
> > choke points that can enforce corporate policy of what an application
> > really should be limited to doing. A for instance would be the
> > limitation of IP access from a DMZ web server to a middle tier server
> > for its data. A firewall between the DMZ server and the middle tier
> > server can be configured to ensure that this is the only access the
> > server has and not access to other systems that it has no reason to be
> > interacting with (such as a mail server). This helps reduce exposure
> > should the DMZ server be compromised. Also, if the firewall between the
> > DMZ server and the middle tier server is configured properly, you will
> > be able to discover irregular activity that could indicate a misbehaving
> > application or a server compromise by traffic trying to conduct actions
> > that were not part of the agreed on policy.
> >
> >
> > On Tue, Mar 25, 2008 at 4:56 AM, william fitzgerald
> > <wfitzgerald (at) tssg (dot) org [email concealed] <mailto:wfitzgerald (at) tssg (dot) org [email concealed]>> wrote:
> >
> > Dear Firewall Experts,
> >
> > Provocative Question:
> > ++++++++++++++++++++
> > Are firewalls obsolete in a world involving enterprise Web Service
> SOA?
> >
> > What do I mean by the above question: given that Web Services (J2EE
> and
> > so forth) tend to tunnel through http and https (eg. SOAP) what role
> can
> > a traditional network firewall play? If its just a matter of opening
> > ports http and https for your dedicated enterprise services then is
> > there even a need for a firewall!
> >
> > I am asking this question not to be flamed but to provoke a
> discussion
> > as to why we still need firewalls.
> >
> > Assumptions:
> > ++++++++++++
> > I use the term firewall loosely to mean "network access control".
> That
> > is, its a mechanism to prevent unwanted packets. Therefore, a
> firewall
> > could be iptables (stateful, DPI etc) or even the proxy TCP
> Wrappers,
> > cisco and so forth.
> >
> > In particular, I have focused on Linux iptables and TCP Wrapper. I
> > realize that one can install an xml based firewall to inspect packet
> > content in regard to web services.
> >
> > Scenario Network:
> > ++++++++++++++++++
> > Internet ---> Firewall ---> Enterprise SOA Server ---> Additional
> > firewalls and back-end database servers etc.
> >
> > Is it a case that in this Enterprise SOA environment the NAC
> firewall is
> > made redundant (as opposed to an xml firewall):
> >
> > Internet ---> Enterprise SOA Webservice server
> >
> > Assuming of course the servers are dedicated Web Service servers
> that
> > run no other services such as DHCP, intranet web server, email and
> so
> > forth that need to be protected?
> >
> > Firewall Justification:
> > +++++++++++++++++++++++
> >
> > I am trying to find publications, white papers, reports etc that
> state
> > the case for the need for firewalls. I need something concrete.
> >
> > The current information I have found (web service orientated!) tends
> to
> > say firewalls are obsolete when talking about enterprise SOA given
> that
> > once port 80 and 443 is open on the firewall the SOS services are
> > exposed and hence protection happens at the application layer of the
> > particular service.
> >
> > However, best practice suggests one should take a more holistic
> approach
> > to security and apply the "belt-and-braces" approach. That is,
> install
> > firewalls, IDS, AV, proper authentication at various OSI stack
> layers
> > etc etc. So we get a layered security affect, thus there must be a
> > justification for using a firewall still.
> >
> > My Opinion:
> > +++++++++++
> >
> > My opinion on what NAC firewalls can offer to web service SOA other
> than
> > simply opening port http and https is as follows:
> >
> > 1) control access to those ports via ip address ranges (eg.
> > customer/business subscribers)
> > 2) deep packet inspection to solicit appropriate content incoming
> and
> > outgoing from the SOA enterprise servers.
> > 3) ???? what else would be done? please comment.
> >
> > While I agree that there are xml based firewalls to monitor xml
> based
> > Web Service traffic, I wonder can it perform access controls at the
> > lower levels like network based firewalls (for example, block
> certain IP
> > addresses)? My guess is they don't given the operate at the
> application
> > layer.
> >
> > I also wonder why one would invest in an xml firewall that is
> dedicated
> > to one kind of traffic profiling and not use for example a very
> > expensive cisco firewall that can cover a multitude of traffic
> > profiling. Presumably these expensive firewalls (or the equivalent
> > unexpensive iptables firewall) can inspect the packet for malicious
> > content to and from the enterprise servers (I believe we have
> > snort-2-iptables to also help here). At any rate, I do not want to
> start
> > a huge debate on the pros and cons of an xml firewall versus a
> network
> > firewall as I am aware dedicated firewalls specialize in various
> traffic
> > profiling. Also its best practice to install a wide range for
> firewall
> > capabilities.
> >
> > The real issue is the justification of NAC's in an enterprise SOA
> > environment. Of course, if this enterprise environment also included
> the
> > company standard services such as email, dns, web server etc I can
> see
> > the major impact of the NAC firewall. But what is the case for
> dedicated
> > enterprise SOA?
> >
> >
> > My shortcomings:
> > ++++++++++++++++
> > My inexperience in an enterprise network environment of how things
> are
> > really carried out rather than what is done in theory.
> >
> >
> > Summary:
> > ++++++++
> > What role do NAC's have to play in an environment of enterprise
> > application services?
> >
> > All pointers to documentation and your comments are welcome.
> >
> > I look forward to your support,
> > regards,
> > Will.
> >
> > --
> > William M. Fitzgerald,
> > PhD Student,
> > Telecommunications Software & Systems Group,
> > ArcLabs Research and Innovation Centre,
> > Waterford Institute of Technology,
> > WIT West Campus,
> > Carriganore,
> > Waterford.
> > Office Ph: +353 51 302937
> > Mobile Ph: +353 87 9527083
> > Web: www.williamfitzgerald.org <http://www.williamfitzgerald.org>
> > www.linkedin.com/in/williamfitzgerald
> > <http://www.linkedin.com/in/williamfitzgerald>
> > www.ryze.com/go/wfitzgerald <http://www.ryze.com/go/wfitzgerald
> >
> >
> >
> >
> >
> >
> >
> > --
> > Kindest Regards,
> >
> > Geoff
>
> --
> William M. Fitzgerald,
> PhD Student,
> Telecommunications Software & Systems Group,
> ArcLabs Research and Innovation Centre,
> Waterford Institute of Technology,
> WIT West Campus,
> Carriganore,
> Waterford.
> Office Ph: +353 51 302937
> Mobile Ph: +353 87 9527083
> Web: www.williamfitzgerald.org
> www.linkedin.com/in/williamfitzgerald
> www.ryze.com/go/wfitzgerald
>
>
>
>
--
Kindest Regards,
Geoff
The answer to your questions about policies is that it requires a joint effort to
establish the policies. You need the input of the developers as to
what it the application should be doing during its normal operation.
The system administrators need to accommodate the developers with a
degree of accessibility to allow the application to function, but never with total access to the system (absolute access is never a good idea). The network administrators configure the network access according to the agreed upon access requirements. However, this process is usually agreed upon during the actual discussion about the application before the code is even written and not at deployment time.<br>
<br>After an application is deployed the admins should be doing what their jobs require them to do: check the logs, patch systems, etc. There's no reason why their roles or responsibilities should need to be changed for a single application. <br>
<br>As for security: firewalls come from everything from the garden variety dumb as bricks models that only feature ip/port restrictions to the more advanced models that use technologies to inspect the traffic before it hits the application and make sure that what is being transmitted falls within parameters that would be considered "normal". The latter models have to understand the protocol being used such as HTTP, FTP, SOAP, etc. SSL can be decrypted by an SSL accelerator applicance making the encryption not a problem for firewall appliances. Application level proxies also can help with security, but require more configuration and are more obvious than the cut-through variety that use SMLI or SPF. <br>
<br>Preventing a total system takeover from a poorly written application requires the usual amount of common sense: the principal of least security (never give anything more privileges than it needs). Different user accounts used for different applications provides a level of segregation between the applications be it for file system or database access.<br>
<br>I'm not really sure if there's really a one stop source on what you're looking for because your questions span three different security categories: network security (covered in network administration books), system security (covered in systems administration books), and designing secure applications (covered in programming books). Then, if there's a database involved (as is usually the case), you have a dba whose job revolves around tweaking the performance and the security of the rdbms (oracle, sql server, db2, etc). The reason for this segregation is usually because the roles in an administration are segregated and that's not necessarilly a bad thing. Generally, the admins are concerned with securing their appliances and systems to the highest state possible while the dba and the programmer are concerned with their respective applications. It's a lot to juggle.<br>
<br>How much lattitude or how many roles anyone has (particularly the developer) is more a matter of the policy that you're working for making it completely arbitrary. There are better practices and guidance available in each field, but knowing everything takes hours and hours of reading plus years of experience. If you're looking at trying your hand in seeing how in depth you can go with programming and security grab some network gear and cheap systems off EBay, construct a lab of your own, and go to town because there's no substitute for doing.<br>
<br><br><div class="gmail_quote">On Wed, Mar 26, 2008 at 10:35 AM, william fitzgerald <<a href="mailto:wfitzgerald (at) tssg (dot) org [email concealed]">wfitzgerald (at) tssg (dot) org [email concealed]</a>> wrote:<br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Thanks Geoffrey,<br>
<br>
I agree firewalls are an important point of demarcation narrowing access<br>
(intended access) between various intranets.<br>
<br>
In your experience what kinds of rules or policies are put in place when<br>
an enterprise application goes live? What are the kinds of policies that<br>
application developers demand from system administrators?<br>
<br>
My argument is that it appears to me that (secure) Enterprise Web<br>
Service applications, particularly those involving access control, are<br>
typically focused at the application-domain only, rather than taking a<br>
more holistic approach to also include the underlying infrastructure<br>
(for example, firewalls). As a result, infrastructure configurations may<br>
unintentionally hinder and prohibit the normal operation of the Web Service.<br>
<br>
Thus, the ideal firewall configuration is one that is aligned with the<br>
application supported by the system, that is, it permits valid<br>
application traffic, and, preferably, no more and no less.<br>
<br>
As I stated in my original post, Web Service developers assume the<br>
underlying infrastructure is automatically available. Also there seems<br>
to be a tendency to tunnel (for example SOAP) over http or https. From<br>
this point of view, Web Service developers may form the opinion that<br>
firewalls are redundant as they typically have ports 80 and 443<br>
accessible (and forward traffic to specialized user-space programs for<br>
further packet processing).<br>
<br>
Maybe this is correct! comments?<br>
<br>
In my opinion, deploying a network level firewall (such as Linux<br>
Netfilter) provisioned for Enterprise Web Services is not simply about<br>
opening port 80 on the server for all traffic; one may wish to deny<br>
certain nodes (IP addresses, etc.), only accept HTTP traffic from some<br>
nodes, require other nodes to use HTTPS and also deal with HTTP traffic<br>
that is tunneled through proxies available on other ports.<br>
<br>
Comments?<br>
<br>
While low level infrastructure such as network firewalls may not solve<br>
all security issues ( as a more suitable application based XML firewall<br>
mite) in regard to Web Service applications, I believe as stated in my<br>
previous post, they have a role to play in applying the belt-and-braces<br>
approach to security best practices.<br>
<br>
Comments?<br>
<br>
What I am really looking for is some concrete documents, publications,<br>
administrator experience that helps clarify the important role of<br>
Network Access Controls (firewalls, IPS etc) within an enterprise SOA<br>
environment, if any.<br>
<br>
regards,<br>
Will.<br>
<div class="Ih2E3d"><br>
Geoffrey Gowey wrote:<br>
> To be succinct: yes, firewalls are still very much a necessity. If one<br>
> ignores the more advanced features of firewalls and only focuses on the<br>
> basic features of ip packet source/destination filtering there's still<br>
> evidence of a clear need for them before even seeing what the usage of<br>
> other technologies such as SMLI or SPF bring to the table. Relying on<br>
> operating system and application security alone is insufficient in a<br>
> networked environment.<br>
><br>
> By putting internal and external firewalls in place, a corporation has<br>
> choke points that can enforce corporate policy of what an application<br>
> really should be limited to doing. A for instance would be the<br>
> limitation of IP access from a DMZ web server to a middle tier server<br>
> for its data. A firewall between the DMZ server and the middle tier<br>
> server can be configured to ensure that this is the only access the<br>
> server has and not access to other systems that it has no reason to be<br>
> interacting with (such as a mail server). This helps reduce exposure<br>
> should the DMZ server be compromised. Also, if the firewall between the<br>
> DMZ server and the middle tier server is configured properly, you will<br>
> be able to discover irregular activity that could indicate a misbehaving<br>
> application or a server compromise by traffic trying to conduct actions<br>
> that were not part of the agreed on policy.<br>
><br>
><br>
> On Tue, Mar 25, 2008 at 4:56 AM, william fitzgerald<br>
</div><div><div></div><div class="Wj3C7c">> <<a href="mailto:wfitzgerald (at) tssg (dot) org [email concealed]">wfitzgerald (at) tssg (dot) org [email concealed]</a> <mailto:<a href="mailto:wfitzgerald (at) tssg (dot) org [email concealed]">wfitzgerald (at) tssg (dot) org [email concealed]</a>>> wrote:<br>
><br>
> Dear Firewall Experts,<br>
><br>
> Provocative Question:<br>
> ++++++++++++++++++++<br>
> Are firewalls obsolete in a world involving enterprise Web Service SOA?<br>
><br>
> What do I mean by the above question: given that Web Services (J2EE and<br>
> so forth) tend to tunnel through http and https (eg. SOAP) what role can<br>
> a traditional network firewall play? If its just a matter of opening<br>
> ports http and https for your dedicated enterprise services then is<br>
> there even a need for a firewall!<br>
><br>
> I am asking this question not to be flamed but to provoke a discussion<br>
> as to why we still need firewalls.<br>
><br>
> Assumptions:<br>
> ++++++++++++<br>
> I use the term firewall loosely to mean "network access control". That<br>
> is, its a mechanism to prevent unwanted packets. Therefore, a firewall<br>
> could be iptables (stateful, DPI etc) or even the proxy TCP Wrappers,<br>
> cisco and so forth.<br>
><br>
> In particular, I have focused on Linux iptables and TCP Wrapper. I<br>
> realize that one can install an xml based firewall to inspect packet<br>
> content in regard to web services.<br>
><br>
> Scenario Network:<br>
> ++++++++++++++++++<br>
> Internet ---> Firewall ---> Enterprise SOA Server ---> Additional<br>
> firewalls and back-end database servers etc.<br>
><br>
> Is it a case that in this Enterprise SOA environment the NAC firewall is<br>
> made redundant (as opposed to an xml firewall):<br>
><br>
> Internet ---> Enterprise SOA Webservice server<br>
><br>
> Assuming of course the servers are dedicated Web Service servers that<br>
> run no other services such as DHCP, intranet web server, email and so<br>
> forth that need to be protected?<br>
><br>
> Firewall Justification:<br>
> +++++++++++++++++++++++<br>
><br>
> I am trying to find publications, white papers, reports etc that state<br>
> the case for the need for firewalls. I need something concrete.<br>
><br>
> The current information I have found (web service orientated!) tends to<br>
> say firewalls are obsolete when talking about enterprise SOA given that<br>
> once port 80 and 443 is open on the firewall the SOS services are<br>
> exposed and hence protection happens at the application layer of the<br>
> particular service.<br>
><br>
> However, best practice suggests one should take a more holistic approach<br>
> to security and apply the "belt-and-braces" approach. That is, install<br>
> firewalls, IDS, AV, proper authentication at various OSI stack layers<br>
> etc etc. So we get a layered security affect, thus there must be a<br>
> justification for using a firewall still.<br>
><br>
> My Opinion:<br>
> +++++++++++<br>
><br>
> My opinion on what NAC firewalls can offer to web service SOA other than<br>
> simply opening port http and https is as follows:<br>
><br>
> 1) control access to those ports via ip address ranges (eg.<br>
> customer/business subscribers)<br>
> 2) deep packet inspection to solicit appropriate content incoming and<br>
> outgoing from the SOA enterprise servers.<br>
> 3) ???? what else would be done? please comment.<br>
><br>
> While I agree that there are xml based firewalls to monitor xml based<br>
> Web Service traffic, I wonder can it perform access controls at the<br>
> lower levels like network based firewalls (for example, block certain IP<br>
> addresses)? My guess is they don't given the operate at the application<br>
> layer.<br>
><br>
> I also wonder why one would invest in an xml firewall that is dedicated<br>
> to one kind of traffic profiling and not use for example a very<br>
> expensive cisco firewall that can cover a multitude of traffic<br>
> profiling. Presumably these expensive firewalls (or the equivalent<br>
> unexpensive iptables firewall) can inspect the packet for malicious<br>
> content to and from the enterprise servers (I believe we have<br>
> snort-2-iptables to also help here). At any rate, I do not want to start<br>
> a huge debate on the pros and cons of an xml firewall versus a network<br>
> firewall as I am aware dedicated firewalls specialize in various traffic<br>
> profiling. Also its best practice to install a wide range for firewall<br>
> capabilities.<br>
><br>
> The real issue is the justification of NAC's in an enterprise SOA<br>
> environment. Of course, if this enterprise environment also included the<br>
> company standard services such as email, dns, web server etc I can see<br>
> the major impact of the NAC firewall. But what is the case for dedicated<br>
> enterprise SOA?<br>
><br>
><br>
> My shortcomings:<br>
> ++++++++++++++++<br>
> My inexperience in an enterprise network environment of how things are<br>
> really carried out rather than what is done in theory.<br>
><br>
><br>
> Summary:<br>
> ++++++++<br>
> What role do NAC's have to play in an environment of enterprise<br>
> application services?<br>
><br>
> All pointers to documentation and your comments are welcome.<br>
><br>
> I look forward to your support,<br>
> regards,<br>
> Will.<br>
><br>
> --<br>
> William M. Fitzgerald,<br>
> PhD Student,<br>
> Telecommunications Software & Systems Group,<br>
> ArcLabs Research and Innovation Centre,<br>
> Waterford Institute of Technology,<br>
> WIT West Campus,<br>
> Carriganore,<br>
> Waterford.<br>
> Office Ph: +353 51 302937<br>
> Mobile Ph: +353 87 9527083<br>
</div></div>> Web: <a href="http://www.williamfitzgerald.org" target="_blank">www.williamfitzgerald.org</a> <<a href="http://www.williamfitzgerald.org" target="_blank">http://www.williamfitzgerald.org</a>><br>
<div class="Ih2E3d">> <a href="http://www.linkedin.com/in/williamfitzgerald" target="_blank">www.linkedin.com/in/williamfitzgerald</a><br>
</div>> <<a href="http://www.linkedin.com/in/williamfitzgerald" target="_blank">http://www.linkedin.com/in/williamfitzgerald</a>><br>
> <a href="http://www.ryze.com/go/wfitzgerald" target="_blank">www.ryze.com/go/wfitzgerald</a> <<a href="http://www.ryze.com/go/wfitzgerald" target="_blank">http://www.ryze.com/go/wfitzgerald</a>><br>
><br>
><br>
><br>
><br>
><br>
><br>
> --<br>
> Kindest Regards,<br>
><br>
> Geoff<br>
<font color="#888888"><br>
--<br>
</font><div><div></div><div class="Wj3C7c">William M. Fitzgerald,<br>
PhD Student,<br>
Telecommunications Software & Systems Group,<br>
ArcLabs Research and Innovation Centre,<br>
Waterford Institute of Technology,<br>
WIT West Campus,<br>
Carriganore,<br>
Waterford.<br>
Office Ph: +353 51 302937<br>
Mobile Ph: +353 87 9527083<br>
Web: <a href="http://www.williamfitzgerald.org" target="_blank">www.williamfitzgerald.org</a><br>
<a href="http://www.linkedin.com/in/williamfitzgerald" target="_blank">www.linkedin.com/in/williamfitzgerald</a><br>
<a href="http://www.ryze.com/go/wfitzgerald" target="_blank">www.ryze.com/go/wfitzgerald</a><br>
<br>
<br>
<br>
</div></div></blockquote></div><br><br clear="all"><br>-- <br>Kindest Regards,<br><br>Geoff
[ reply ]