Quote:
Originally Posted by Wind_mill In a multi-layered secure environment , throughput of the firewall matters while designing Internal Firewalls as well. A low throughput may create a bottleneck otherwise. We have architected soultions for couple of major Banks here in Europe and I can share some practical experiences in this.
I'm out of core technology roles for last couple of years and now part of a Big-4 consulting firm. However I thought I need to share the following experiences with Pix. (Things might have changed with ASA).
@Samurai:
I'm not sure about the existing architecture of the network where you are planning to incorporate Pix/ASA. Considering that you don't have a fulltime Network Engineer ,I suggest you to get a demo box and try it before actually buying/deploying the same.
We did a migration from Checkpoint NG to 515E back in 2002. Had to do lot of debugging to resolve un-foreseen issues.
Issue #1
We had a Oracle 8i server sitting behind Checkpoint Firewall with ODBC/JDBC Clients or App Servers connecting to it from other part of the world over VPN links. This just stopped working after PIX implementation. While Cisco TAC was unable to find the exact issue , our Oracle Support partner confirmed that its due to a compatibility isuue between SQLNET and CISCO NAT . We had to change the design of database connectivity - We used Oracle connection manager just to get compatible with Pix and had to role out this change in all locations. You can still find the threads related to this issue - just search Pix+NAT+Oracle in google.
Issue #2
The IP based video conferencing equipment (Polycom) stopped functioning . This was working without any issue earlier with Checkpoint NG . Cisco TAC confirmed that its due to the way Pix is handling H323 traffic . Finally they advice to remove the "Fix-up" of the same protocol and resolved it- this means firewall intelligence over this traffic is disabled. I strongly suggest to keep the Vonage device outside the firewall to avoid such issues . I don't think there are much security issues in doing so.
Issue#3
We had a Web server kept in India and Client Applications (Developed in VC++) deployed across the US connecting to it using SOAP service. Pix way of handling Http traffic created a bottleneck here and nobody could "POST" any data. This was again resolved at the cost of disabling fix-up http or turning off the intelligent feature. To me , once it is turned off what we get is mere port level security.
Hope this helps! |
@windmill - let me clarify
1) we are talking of a perimeter firewall here in the case of samurai - not an internal firewall. Any outsider accessing his webservers behind the firewall will not go faster than his isp link which I don't think is more than 150mbps (samurai please correct me). Now I am sure he will have insiders (read lan users) accessing the webservers on the firewall dmz and that is where if and only if they are accessing at gigabit speeds (loads of users streaming video) then there will be a bottleneck. I am assuming this is not the case.
2) Things have changed a lot with both the PIXes and the ASAs. For one the PIXes when running fixups were on code 6.x. This was replaced with 7.x with application inspection feature, which included support for over 30 protocols (including http, voice, video, sip, skinny, rtp, h.323, h.243 etc) adn this was certified as part of CC EAL4+ testing. This was further improved in 8.x release and both 7.x and 8.x run on PIX and ASA. PIX is now EOLed and ASA is the default offering.
3) Checkpoint was the first to introduce h.323, but as of today it cannot be compared to Cisco ASAs voice/video support (which is derived from Cisco's experience in the VOIP and video industry - something that Checkpoint will never have). Checkpoint's voice/video inspection (nor any other vendors) is not part of the security evaluation target on their CC EAL4+ cert, whereas Cisco's is (links posted earlier). Supported protocol for voice video are sip. skinny, h.323 v1-4, gtp (3g mobile wireless), mgcp, rtp, rtcp, rtsp, tapi, jtapi
4) I am aware of the Oracle issue, I believe this was faced with 6.x code. This may have been resolved in the 7.x release when app inspection support for databsed was introduced to dynamically inspect and open up ports for ils/ldap, oracle sql *net (v1/2), ms rpc/dce rpc, ms n/wing, nfs, rsh, sunrcp, nis+, xwindows (xdmcp)
5) there are currently no known issues with video conferencing over IP as long as it runs over standards (read rfc non proprietry) based protocols. A lot of other solutions including Cisco's own video conferencing over IP and Telepresence solutions (featured in ads these days) work with ASAs. ASAs are tested thoroughly in voice/video environments by Cisco's own voice technonology business group
6) Among the other protocols in app inspection there is a very detailed http inspection map, which I have used successfully in many implementations to handle everything from basic URI request handling, http command filtering to dropping cross site scripting (XSS) attacks. I am not an expert at SOAP, but I am sure release 7.x onwards it can be easily handled.
7) The overall point I am trying to make here yes PIX fixups were a mess, I hated them and I was always a fan of checkpoints app inspection. However Cisco in the last few years has vastly progressed past those fixup days and is now comparable or better (in some cases like voip or video over ip) to other vendors in app inspection.
8) did i just make another long post !!
9) phew
