Important note about SSL VPN compatibility for 20.0 MR1 with EoL SFOS versions and UTM9 OS. Learn more in the release notes.

This discussion has been locked.
You can no longer post new replies to this discussion. If you have a question you can start a new discussion

Looking for a "best practice" design or tips for multiple location/firewall authentication

Hi,

we have a customer, who has about 10 branch offices with each 5 to 50 users and a headquarter with about 50 users.
Every BO has its own XGS firewall, which is currently connected via IPSEC VPN and will later be connected via MPLS to the HQ.
In HQ there is a Citrix terminal server farm with about 25 app servers. Most applications the users use come from Citrix, some local applications exist because they aren't terminal server compatible.

Internet breakout is only via HQ firewall (local networks to Any policy based VPNs on the BO firewalls).
Every client and server has Sophos Endpoint/Server Protection installed.
Infrastructure is centralized (no servers in BO, BO firewalls are acting as DHCP relays.




Currently only the HQ firewall is doing authentication. HQ's Sophos endpoints are authenticating automatically via heartbeat, the TS sessions via SATC. STAS is currently not in use.

I am looking for a way to optimize this because the BO clients for example are currently not authenticated automatically via heartbeat, only the HQ IP addresses show up for "Live users". So only the Citrix applications part for BO users is covered via authentication, but not the local applications part,

Any suggestions how to make this more admin friendly/secure?
My goal is to authenticate local AND remote users on the HQ firewall. Authenticating the users on the BO firewall would only be a bonus. *)

If more detailed informations are required feel free to ask!

Regards,

Kevin

*) Unfortunately the customer historically has 9 different sub-companies, each with it's own mail TLD.
Therefore I have 10 different authentication servers defined on the HQ's XG cluster, 1 for general Active Directory, plus another for each mail TLD, because otherwise the HB authentication didn't work properly.



This thread was automatically locked due to age.
  • One thing to add:

    Because of the authentication differencies (SATC via AD, heartbeat via own entry) I have each user (at least the HQ users) with a second mail address/firewall user account. One corresponds to his windows UPN login (heartbeat) one corresponds to the internal AD domain which is different from any "real" mail address (SATC).

    If I could get rid of this I would be very obliged.

    Regards,

    Kevin

    Sophos CE/CA (XG, UTM, Central Endpoint)
    Gold Partner

  • Hi kerobra 

    Try to move from policy base VPN to route base VPN try route-based VPN (any to any subnets) might help 

    Regards

    "Sophos Partner: Networkkings Pvt Ltd".

    If a post solves your question please use the 'Verify Answer' button.

  • Unfortunately, Heartbeat and Authentication do not sync over multiple Firewalls. I never understood why this is not possible. I mean, they have Central. It would not be too much rocket science to use that to Sync User and HB all over your distributed networks.

     Heartbeat accross multiple Firewalls - somehow possible? 

    Now you must decide where you want HB to happen - you have chosen the HQ firewall.

    What you can do, is to bring the heartbeat IP connection from the BO to the HQ. That way only the HQ firewall will handle all authentication and HB information.So you do not want the BO Firewall to compute the heartbeat request and send it to Central, instead you want that to be forwarded to the HQ firewall and that device syncs all heartbeat sessions with Central.

    As far as you do not have the need for HB-based Firewall rules at the BO, this will fulfill your need. Route based VPN with "Tunnel interface" can be used to do SD-WAN routing for heartbeat IP and Port.

  • Unfortunately this is no option because the DHCP scopes are managed centrally on a MS Windows DHCP server. I wanted to use tunnel interfaces first unless I stumbled over this entry in the documentation: (3rd row)

    Because DHCP is generally managed by our 1st level support which absolutely would be overextended by doing this on a firewall (and apart from that I don't want them to be able to look into firewalls) it is no option to terminate DHCP on every BO firewall. Disappointed

    Regards,

    Kevin

    Sophos CE/CA (XG, UTM, Central Endpoint)
    Gold Partner

  • I understand. have you thought of DNAT the HB IP/Port as sugegsted by in that post I linked above?

    Because of the authentication differencies (SATC via AD, heartbeat via own entry) I have each user (at least the HQ users)

    We're using that combination and have no issues with duplicate usernames - both authentications show (when SATC even works at all) the correct username on the firewall - it is the UPN (not really the mail address).

  • "Sophos Partner: Networkkings Pvt Ltd".

    If a post solves your question please use the 'Verify Answer' button.

  • The UPN at our customer is identical to his mail address since they need it for Entra ID synchronization.
    Their classic SAMAccount name is different to the local part of the UPN.

    For example: SAMAccount is "jsmith", UPN is john.smith@external-domain.tld, internal AD domain is "lan.external-domain.tld".
    Because of the weird behaviour how "Synchronized user ID authentication" is working ("The domain is taken from the User Principle Name (UPN) of the users' AD record, and the username is taken from the sAMAccountName") this leads to multiple accounts at the firewall.

    1st account: jsmith@external-domain.tld
    2
    nd account: jsmith@lan.external-domain.tld.

    Both firewall accounts have nothing to to with the "real UPN" or mail address and both accounts have the same mail address:
    Fortunately the HQ firewall is not used for antispam services and it is not used for client VPN. If the users had to go into the Userportal it would get very confusing.

    Regards,

    Kevin

    Sophos CE/CA (XG, UTM, Central Endpoint)
    Gold Partner

  • Don't hang on the underlying VPN. The whole IPSEC tunnel topic will soon die (app. 3-6 months) because of moving from local WANs to MPLS in every BO location.

    I have one BO firewall that is currently connected via "Etherconnect" which is a technique similar to a RED tunnel. To cut it easy, you have a RJ45 GBit Port in HQ and a RJ45 GBit Port in BO and they share the same L2/3 like it would be with a cable between two switches.

    Maybe I will use this firewall for the testing of the DNAT thing for HB as it is nearly the state that the current IPSEC connected BO firewalls will have in a few months.

    Regards,

    Kevin

    Sophos CE/CA (XG, UTM, Central Endpoint)
    Gold Partner