[cisco-voip] Expressway E Firewall Rule Activation

Tue Apr 30 13:28:15 EDT 2019

If you stick the Edge LAN1 on the inside (where Core is), then doesn't this
technically circumvent the "traversal" part of the technology?  Because we
point Core at Edge via it's LAN 1 IP, right?  Or am I missing something?

On Tue, Apr 30, 2019 at 11:33 AM Ryan Huff <ryanhuff at outlook.com> wrote:

> Not generally, no. A couple of my larger customer’s that have fully
> fleshed out IT departments did though.
>
> For a few of my customers I’ve had to walk them through setting a 2nd one
> up. In some cases, not even a true DMZ and just a new network and lock it
> down with ACLs.
>
> I’ve also had customer’s which do the DMZ on “LAN2” (outside), and then
> keeps LAN1 in the same network as Expressway-C. This particular method
> doesn’t offer a lot of advantages (from a infosec perspective) over a
> “Single NIC”, but still makes the traffic flow more logical, easier to
> support and troubleshoot and keeps you from having to “hairpin” in the
> firewall (ewww, like gag me with a spoon man lol), which I have never been
> a fan of from a design perspective.
>
> -Ryan
>
> On Apr 30, 2019, at 12:12, Anthony Holloway <
> avholloway+cisco-voip at gmail.com> wrote:
>
> Ryan,
>
> Do you have any insight as to whether or not it's common for Firewalls in
> the field to already have more than one DMZ defined?  In my limited
> experience, I have never seen it done, and I am having to have that second
> DMZ created to support Expressway.  For that reason, I actually tend to
> think the single NIC approach is better, although, the NAT reflection could
> be a limitation of some firewalls.
>
> On Tue, Apr 30, 2019 at 11:09 AM Ryan Huff <ryanhuff at outlook.com> wrote:
>
>> Adam,
>>
>> I certainly didn't mean to imply the, "Expressway Edge on a Stick" method
>> doesn't work, though out of pure technical curiosity, I would be curious as
>> to what exists in your environment that would make a " single NIC"
>> Expressway Edge deployment more preferred than "dual NICs" (not that I
>> expect you would or could say). I can think of very few reasons that a
>> single NIC edge would be more ideal than a dual NIC edge (outside of the
>> infosec team just not wanting to screw with the firewall, or production not
>> being able to sustain a maintenance window); its easier to troubleshoot,
>> easier to install, easier to support and easier to secure.
>>
>> Though, I suspect I'm, "preaching to the choir", lol 😉. All good my
>> friend.
>>
>> Thanks,
>>
>> Ryan
>>
>> ------------------------------
>> *From:* Pawlowski, Adam <ajp26 at buffalo.edu>
>> *Sent:* Tuesday, April 30, 2019 11:36 AM
>> *To:* 'Ryan Huff'
>> *Cc:* cisco-voip at puck.nether.net
>> *Subject:* RE: [cisco-voip] Expressway E Firewall Rule Activation
>>
>>
>> Ryan,
>>
>>
>>
>> The “tl;dr” is that we were sort of given the recommendation by Cisco to
>> just run it with the single interface given our environment and
>> requirements, and hasn’t given us any trouble that I can recall.
>>
>>
>>
>> Long story is …
>>
>>
>> Our environment ends up being the driver for a lot of this, as it is sort
>> of a historic design from the early internet, with just about everything on
>> public address space, and various services and networks secured behind
>> firewalls as needed from internal and external alike.
>>
>>
>>
>> In the dual interface design, the outside interface sits in a “DMZ” with
>> a firewall, which we don’t have available explicitly. There is a border
>> firewall but that isn’t really its function. The inside leg has to sit
>> somewhere as well, which is a place that doesn’t exist.
>>
>>
>> We did have a competitor’s border proxy become compromised in the past
>> due to a software update, and this model where the inside wasn’t properly
>> secured – and given our current VMWare topology, creating another zone to
>> hairpin traffic around to separate that inside interface wasn’t in the
>> cards. Not to mention the annoyance of trying to setup split routes on this
>> device to allow some traffic to go in, some to go out, in an environment
>> that is MRA only.
>>
>>
>>
>> If you trust the E enough never to be a bad actor, then you could put
>> that interface in the same zone as your other collaboration appliances,
>> like the Expressway C, but, we didn’t want to do that either really.
>>
>>
>>
>> Given that, we did have a call with Cisco to discuss this, and with
>> representation from the Expressway group they recommended that we stick
>> with the single interface design.  That was based on the public addressing
>> (so we could avoid NAT reflection) and that despite the pipe dream of
>> everyone wanting HD video calling and mobile client access, we didn’t see
>> that we’d be pushing that much traffic.
>>
>>
>>
>> As it is, the E clusters sit in a collaboration DMZ, where they are
>> independent from any of our other appliances and treated like any other
>> host on our network. Our application firewalls do not allow anything in
>> from the Expressway E since the C tunnels to it, so really the only thing
>> lacking from a security standpoint there could be containment of that host,
>> but, we chose to guard from it instead.
>>
>>
>>
>> Since we installed it back on X8.8 or whatever, I’d noted that rebooting
>> the appliance does not reapply the internal rules, which can easily be
>> forgotten, and would need to be remembered if you run a VMWare HA policy
>> that restarts the guest.
>>
>>
>>
>> That all being said the worst that we have seen are various SSH attempts
>> (on any port, the zone tunnel, administrative SSH, doesn’t matter) until
>> the rules are put back up. We could tighten them on the border once that
>> becomes available to do so.
>>
>>
>>
>> The B2BUA is invoked on calls within the appliances sometimes which can
>> cause some confusion with attempting to read logging if need be, but it
>> hasn’t otherwise caused us any trouble.
>>
>>
>>
>> Adam
>>
>>
>>
>>
>>
>>
>>
>> *From:* Ryan Huff <ryanhuff at outlook.com>
>> *Sent:* Tuesday, April 30, 2019 10:13 AM
>> *To:* Pawlowski, Adam <ajp26 at buffalo.edu>
>> *Cc:* cisco-voip at puck.nether.net
>> *Subject:* Re: [cisco-voip] Expressway E Firewall Rule Activation
>>
>>
>>
>> That seems odd and not been my experience. Let me ask; why are you using
>> the application firewall rather than the actual firewall (another reason
>> all our edge’s should be using dual interfaces with LAN1 and LAN2 in their
>> own separate security zones)? Is there a reason you have to, in other words?
>>
>> Thanks,
>>
>>
>>
>> Ryan
>>
>>
>> On Apr 30, 2019, at 08:49, Pawlowski, Adam <ajp26 at buffalo.edu> wrote:
>>
>> Figured I’d also ask this question
>>
>>
>>
>> I note that it seems like any time I reboot an Expressway E, I have to go
>> and re-activate all the firewall rules. They don’t seem to activate
>> automatically.
>>
>>
>>
>> Is there something I missed or is this really what’s necessary?
>>
>>
>>
>> Adam
>>
>>
>>
>>
>>
>> _______________________________________________
>> cisco-voip mailing list
>> cisco-voip at puck.nether.net
>>
>> https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fpuck.nether.net%2Fmailman%2Flistinfo%2Fcisco-voip&data=02%7C01%7C%7C3fcc9eb351fe41b70dfc08d6cd6a4a65%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C636922253726465693&sdata=72kYzwChhoFD14H6a6mRTn4TdHUcMDcFWrMSXpRo%2Btw%3D&reserved=0
>> <https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fpuck.nether.net%2Fmailman%2Flistinfo%2Fcisco-voip&data=02%7C01%7C%7Cc87fe8db81b44e792e1a08d6cd8694bd%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C636922375231674202&sdata=Kas96cIptwcxYdyXhvByyrhzp9V6ukCBtNP%2B07XoQ10%3D&reserved=0>
>>
>> _______________________________________________
>> cisco-voip mailing list
>> cisco-voip at puck.nether.net
>> https://puck.nether.net/mailman/listinfo/cisco-voip
>> <https://eur02.safelinks.protection.outlook.com/?url=https%3A%2F%2Fpuck.nether.net%2Fmailman%2Flistinfo%2Fcisco-voip&data=02%7C01%7C%7Cc87fe8db81b44e792e1a08d6cd8694bd%7C84df9e7fe9f640afb435aaaaaaaaaaaa%7C1%7C0%7C636922375231694217&sdata=k6IoF7u0lFqFswflMi3gO1wooYbIs0S5LLcRqDFL7Cw%3D&reserved=0>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://puck.nether.net/pipermail/cisco-voip/attachments/20190430/5f21738d/attachment.html>