Re: Outgoing IPSEC

From: Securi Net (securinet2004_at_yahoo.ca)
Date: 11/22/05

  • Next message: mickael kael: "Re: Remote Control to a remote computer with no listen port"
    Date: Tue, 22 Nov 2005 11:29:10 -0500 (EST)
    To: Jason Thompson <securitux@gmail.com>
    
    

    Thanks again, Jason.

    That clarifies it and gives me a strong case in
    denying such requests.

    Thank you everyone for your responses.

    Regards

    CP

    --- Jason Thompson <securitux@gmail.com> wrote:

    > No problem.
    >
    > With a stateful inspection device (firewall), it
    > does allow
    > bidirectional traffic as long as your client
    > initiates the VPN
    > connection to the endpoint. Once the client connects
    > to the endpoint
    > it creates a connection that your firewall
    > recognizes as 'in state'
    > and allows traffic between the two devices. This is
    > by design. So with
    > a stateful inspection firewall, even though you are
    > creating rules
    > that say allow only outbound from client to
    > endpoint, you are
    > essentially saying 'only the client is allowed to
    > initiate a
    > connection'. Once the client creates the tunnel,
    > bidirectional traffic
    > is permitted so long as it obeys the rules of IP
    > "state" (same IP
    > addresses, connection kept alive, in the case of
    > TCP: syn, syn/ack,
    > ack, push, push/ack, sequence numbers,
    > acknowledgement numbers, etc).
    > UDP state is also kept using timeouts, since UDP is
    > a stateless
    > protocol.
    >
    > Keep in mind, the ACTUAL TCP, UDP, and IP header (of
    > a tunnelled
    > connection) from a client or VPN endpoint is
    > encrypted / encapsulated
    > with a completely different header. So an IKE UDP
    > packet's
    > encapsulated data from the endpoint to the client
    > (which is in state)
    > may be a TCP SYN (beginning of a TCP connection from
    > the other network
    > to the client).
    >
    > -J
    >
    > On 11/21/05, Securi Net <securinet2004@yahoo.ca>
    > wrote:
    > >
    > > Jason,
    > >
    > > Thank you for your elaborate response to my query.
    > >
    > > Excuse my ignorance of the way IPSEC tunnels are
    > > established here, but by permitting only outgoing
    > > traffic on port 500, would I not succeed in
    > forcing a
    > > uni-directional tunnel to the external
    > organization.
    > > Or is my understanding of the technology off the
    > mark.
    > >
    > > We have mulled over the option of forcing him onto
    > a
    > > seprate VLAN, but need a clinical argument about
    > the
    > > secuirty risks in opening up IPSEC for him. This
    > is
    > > also being backed by an acceptable use policy.
    > >
    > > Thanks again for your responses.
    > >
    > > CP
    > >
    > > --- Jason Thompson <securitux@gmail.com> wrote:
    > >
    > > > Here's one of the big issues which deals with
    > the
    > > > endpoint rather than
    > > > Internet threats.
    > > >
    > > > Once the VPN connection is established, you have
    > no
    > > > control over what
    > > > traffic is transmitted on the tunnelled network.
    > You
    > > > essentially open
    > > > an unchecked bidirectional link between the VPN
    > > > client and the network
    > > > to which he is connecting on the other end. You
    > are
    > > > essentially
    > > > relying on the other company to enforce a policy
    > on
    > > > that contractor,
    > > > which is a nauseating thought. Also in the case
    > of
    > > > a split tunnel,
    > > > that contractor's PC could be used by someone or
    > > > something in the
    > > > other organization as a jumping point into your
    > > > network. Further to
    > > > that, since the traffic is encrypted, you do not
    > > > have the ability to
    > > > monitor what the contractor is doing. This
    > > > individual could be doing
    > > > anything from browsing questionable sites
    > through
    > > > company X's network
    > > > to receiving the worm-du-jour. Single tunnel
    > does
    > > > nothing here,
    > > > because in a single tunnel situation once
    > > > disconnected from company
    > > > X's network, it will spread throughout yours.
    > > >
    > > > It's all about acceptable risk of course, and
    > > > unfortunately in a lot
    > > > of cases contractor access to VPN is a
    > requirement,
    > > > so do what you can
    > > > to lock it down. First off put the contractor on
    > a
    > > > different VLAN than
    > > > other users and do not allow access to your
    > internal
    > > > resources; the
    > > > VLAN should route right to your firewall. If
    > s/he
    > > > needs access to
    > > > internal resources, then s/he does it with one
    > of
    > > > your company's PC's.
    > > > Also, make the contractor and the other
    > organization
    > > > sign an
    > > > acceptable use policy as well as a policy
    > specifying
    > > > that 'due care'
    > > > will be taken while operating inside the network
    > (AV
    > > > always on,
    > > > desktop firewall, regular AV scans and updates,
    > > > etc). That way if they
    > > > introduce something nasty into your environment
    > and
    > > > they didn't
    > > > exercise due care, they can be held liable. But
    > most
    > > > importantly, the
    > > > policy shows the other organization that you
    > take
    > > > information security
    > > > seriously and you have your eye on them.
    > > >
    > > > If s/he needs access just to get e-mail, the
    > > > consulting company should
    > > > be putting up an OWA server with two factor
    > auth...
    > > > if they don't have
    > > > one already. Some companies require it as they
    > don't
    > > > allow VPN out at
    > > > all.
    > > >
    > > > -J
    > > >
    > > > On 11/18/05, Securi Net <securinet2004@yahoo.ca>
    > > > wrote:
    > > > > Hello List members,
    > > > >
    > > > > I have a question on risks associated with
    > > > allowing
    > > > > outgoing IPSEC traffic on a firewall.
    > > > >
    > > > > I have a contractor who works onsite within
    > our
    > > > > network and needs outgoing port 500 opened on
    > our
    > > > > firewall for him to vpn into his company
    > network.
    > > > >
    > > > > I would like to know about the risks involved
    > in
    > > > > facilitating such access outside, as I have
    > heard
    > > > some
    > > > > talk about security issues around split
    > > > tunnelling. As
    > > > > far as I can understand it, the only threat to
    > our
    > > > > network from the outside would be if someone
    > on
    > > > the
    > > > > outside tries to spoof a session inside using
    > an
    > > > > existing outward connection.
    > > > >
    > > > > Can anyone shed some light on what I shud be
    > > > concerned
    > > > > about here.
    > > > >
    > > > > CP
    >
    === message truncated ===

            

            
                    
    __________________________________________________________
    Find your next car at http://autos.yahoo.ca


  • Next message: mickael kael: "Re: Remote Control to a remote computer with no listen port"

    Relevant Pages

    • Re: Outgoing IPSEC
      ... bidirectional traffic as long as your client initiates the VPN ... Once the client connects to the endpoint ... it creates a connection that your firewall recognizes as 'in state' ...
      (Security-Basics)
    • [PATCH 0/5] [RFC] AF_RXRPC socket family implementation [try #3]
      ... These patches together supply secure client-side RxRPC connectivity as a Linux ... kernel socket family. ... presentation side is left to the client. ... Each connection goes to a particular "service". ...
      (Linux-Kernel)
    • [PATCH 0/5] [RFC] AF_RXRPC socket family implementation
      ... These patches together supply secure client-side RxRPC connectivity as a Linux ... Make it possible for the client socket to be used to go to more than one ... Each connection goes to a particular "service". ...
      (Linux-Kernel)
    • [PATCH 0/5] [RFC] AF_RXRPC socket family implementation [try #2]
      ... These patches together supply secure client-side RxRPC connectivity as a Linux ... Make it possible for the client socket to be used to go to more than one ... Each connection goes to a particular "service". ...
      (Linux-Kernel)
    • Re: .Net Scalability problem
      ... LoadRunner will peak out a server with a few virtual users. ... To get an idea of load, ... Fire off the test client and watch the number of ... > So I think that the MTC generate concurrent connection and per ...
      (microsoft.public.dotnet.framework.adonet)