Do most of you have the phones authenticate incoming calls? We haven't been, but occasionally find a router that has unfiltered full cone NAT (Cisco) or that puts one phone on 5060 with no filtering by IP. The result is that the phone will start ringing at random as script kiddies hit the IP and port 5060 trying to find servers to exploit. I don't see a downside to changing to auth, but not having done it outside of a few tests of a small number of phones, I figured I would ask.
That would depend on the sending UA's ability to answer such challenges, unless you're referring to some setting to white-list IPs or restrict sources to previously called endpoints. On August 8, 2018 1:43:45 PM EDT, Carlos Alvarez <caalvarez at gmail.com> wrote:
Do most of you have the phones authenticate incoming calls? We haven't been, but occasionally find a router that has unfiltered full cone NAT (Cisco) or that puts one phone on 5060 with no filtering by IP. The result is that the phone will start ringing at random as script kiddies hit the IP and port 5060 trying to find servers to exploit. I don't see a downside to changing to auth, but not having done it outside of a few tests of a small number of phones, I figured I would ask.
-- Alex -- Sent via mobile, please forgive typos and brevity.
Using TCP or TLS would avoid open NAT issue, and can cure some naughty SIP ALG issues as well, assuming you want to tolerate the overhead. For UDP, we've used both Digest and Source request validation with Polycom devices. Source validation is probably the easiest route, assuming the UA doesn't need to receive calls from anyone but its proxy or registrar. Digest (nonce challenge) is better if you want to accept calls from anyone who knows your password, but we had an issue with a softswitch that would properly handle auth channel to INVITE but choked when a BYE was challenged. Regards, *Calvin Ellison* Voice Operations Engineer calvin.ellison at voxox.com +1 (213) 285-0555 ----------------------------------------------- *voxox.com <http://www.voxox.com/> * 5825 Oberlin Drive, Suite 5 San Diego, CA 92121 [image: Voxox] On Wed, Aug 8, 2018 at 10:43 AM, Carlos Alvarez <caalvarez at gmail.com> wrote:
Do most of you have the phones authenticate incoming calls? We haven't been, but occasionally find a router that has unfiltered full cone NAT (Cisco) or that puts one phone on 5060 with no filtering by IP. The result is that the phone will start ringing at random as script kiddies hit the IP and port 5060 trying to find servers to exploit. I don't see a downside to changing to auth, but not having done it outside of a few tests of a small number of phones, I figured I would ask.
_______________________________________________ VoiceOps mailing list VoiceOps at voiceops.org https://puck.nether.net/mailman/listinfo/voiceops
I would have to agree with Calvin. Just use TCP. On August 8, 2018 1:58:47 PM EDT, Calvin Ellison <calvin.ellison at voxox.com> wrote:
Using TCP or TLS would avoid open NAT issue, and can cure some naughty SIP ALG issues as well, assuming you want to tolerate the overhead.
For UDP, we've used both Digest and Source request validation with Polycom devices. Source validation is probably the easiest route, assuming the UA doesn't need to receive calls from anyone but its proxy or registrar. Digest (nonce challenge) is better if you want to accept calls from anyone who knows your password, but we had an issue with a softswitch that would properly handle auth channel to INVITE but choked when a BYE was challenged.
Regards,
*Calvin Ellison* Voice Operations Engineer calvin.ellison at voxox.com +1 (213) 285-0555
----------------------------------------------- *voxox.com <http://www.voxox.com/> * 5825 Oberlin Drive, Suite 5 San Diego, CA 92121 [image: Voxox]
On Wed, Aug 8, 2018 at 10:43 AM, Carlos Alvarez <caalvarez at gmail.com> wrote:
Do most of you have the phones authenticate incoming calls? We haven't been, but occasionally find a router that has unfiltered full cone NAT (Cisco) or that puts one phone on 5060 with no filtering by IP. The result is that the phone will start ringing at random as script kiddies hit the IP and port 5060 trying to find servers to exploit. I don't see a downside to changing to auth, but not having done it outside of a few tests of a small number of phones, I figured I would ask.
_______________________________________________ VoiceOps mailing list VoiceOps at voiceops.org https://puck.nether.net/mailman/listinfo/voiceops
-- Alex -- Sent via mobile, please forgive typos and brevity.
That's a change I've never investigated. Or more precisely, haven't investigated since the days when the advice for doing it was "good luck!!" On Wed, Aug 8, 2018 at 11:00 AM Alex Balashov <abalashov at evaristesys.com> wrote:
I would have to agree with Calvin. Just use TCP.
On August 8, 2018 1:58:47 PM EDT, Calvin Ellison <calvin.ellison at voxox.com> wrote:
Using TCP or TLS would avoid open NAT issue, and can cure some naughty SIP ALG issues as well, assuming you want to tolerate the overhead.
For UDP, we've used both Digest and Source request validation with Polycom devices. Source validation is probably the easiest route, assuming the UA doesn't need to receive calls from anyone but its proxy or registrar. Digest (nonce challenge) is better if you want to accept calls from anyone who knows your password, but we had an issue with a softswitch that would properly handle auth channel to INVITE but choked when a BYE was challenged.
Regards,
*Calvin Ellison* Voice Operations Engineer calvin.ellison at voxox.com +1 (213) 285-0555
----------------------------------------------- *voxox.com <http://www.voxox.com/> * 5825 Oberlin Drive, Suite 5 San Diego, CA 92121 [image: Voxox]
On Wed, Aug 8, 2018 at 10:43 AM, Carlos Alvarez <caalvarez at gmail.com> wrote:
Do most of you have the phones authenticate incoming calls? We haven't been, but occasionally find a router that has unfiltered full cone NAT (Cisco) or that puts one phone on 5060 with no filtering by IP. The result is that the phone will start ringing at random as script kiddies hit the IP and port 5060 trying to find servers to exploit. I don't see a downside to changing to auth, but not having done it outside of a few tests of a small number of phones, I figured I would ask.
_______________________________________________ VoiceOps mailing list VoiceOps at voiceops.org https://puck.nether.net/mailman/listinfo/voiceops
-- Alex
-- Sent via mobile, please forgive typos and brevity. _______________________________________________ VoiceOps mailing list VoiceOps at voiceops.org https://puck.nether.net/mailman/listinfo/voiceops
That has changed greatly since 2005. On August 8, 2018 2:07:50 PM EDT, Carlos Alvarez <caalvarez at gmail.com> wrote:
That's a change I've never investigated. Or more precisely, haven't investigated since the days when the advice for doing it was "good luck!!"
On Wed, Aug 8, 2018 at 11:00 AM Alex Balashov <abalashov at evaristesys.com> wrote:
I would have to agree with Calvin. Just use TCP.
On August 8, 2018 1:58:47 PM EDT, Calvin Ellison <calvin.ellison at voxox.com> wrote:
Using TCP or TLS would avoid open NAT issue, and can cure some naughty SIP ALG issues as well, assuming you want to tolerate the overhead.
For UDP, we've used both Digest and Source request validation with Polycom devices. Source validation is probably the easiest route, assuming the UA doesn't need to receive calls from anyone but its proxy or registrar. Digest (nonce challenge) is better if you want to accept calls from anyone who knows your password, but we had an issue with a softswitch that would properly handle auth channel to INVITE but choked when a BYE was challenged.
Regards,
*Calvin Ellison* Voice Operations Engineer calvin.ellison at voxox.com +1 (213) 285-0555
----------------------------------------------- *voxox.com <http://www.voxox.com/> * 5825 Oberlin Drive, Suite 5 San Diego, CA 92121 [image: Voxox]
On Wed, Aug 8, 2018 at 10:43 AM, Carlos Alvarez <caalvarez at gmail.com> wrote:
Do most of you have the phones authenticate incoming calls? We haven't been, but occasionally find a router that has unfiltered full cone NAT (Cisco) or that puts one phone on 5060 with no filtering by IP. The result is that the phone will start ringing at random as script kiddies hit the IP and port 5060 trying to find servers to exploit. I don't see a downside to changing to auth, but not having done it outside of a few tests of a small number of phones, I figured I would ask.
_______________________________________________ VoiceOps mailing list VoiceOps at voiceops.org https://puck.nether.net/mailman/listinfo/voiceops
-- Alex
-- Sent via mobile, please forgive typos and brevity. _______________________________________________ VoiceOps mailing list VoiceOps at voiceops.org https://puck.nether.net/mailman/listinfo/voiceops
-- Alex -- Sent via mobile, please forgive typos and brevity.
It has, but it wasn't that long ago that people were still having challenges. Our preferred phone vendor, Grandstream, still generally advises against it. So...who else on the list uses TCP and has any comments about it? On Wed, Aug 8, 2018 at 11:12 AM Alex Balashov <abalashov at evaristesys.com> wrote:
That has changed greatly since 2005.
On August 8, 2018 2:07:50 PM EDT, Carlos Alvarez <caalvarez at gmail.com> wrote:
That's a change I've never investigated. Or more precisely, haven't investigated since the days when the advice for doing it was "good luck!!"
On Wed, Aug 8, 2018 at 11:00 AM Alex Balashov <abalashov at evaristesys.com> wrote:
I would have to agree with Calvin. Just use TCP.
On August 8, 2018 1:58:47 PM EDT, Calvin Ellison <calvin.ellison at voxox.com> wrote:
Using TCP or TLS would avoid open NAT issue, and can cure some naughty SIP ALG issues as well, assuming you want to tolerate the overhead.
For UDP, we've used both Digest and Source request validation with Polycom devices. Source validation is probably the easiest route, assuming the UA doesn't need to receive calls from anyone but its proxy or registrar. Digest (nonce challenge) is better if you want to accept calls from anyone who knows your password, but we had an issue with a softswitch that would properly handle auth channel to INVITE but choked when a BYE was challenged.
Regards,
*Calvin Ellison* Voice Operations Engineer calvin.ellison at voxox.com +1 (213) 285-0555
----------------------------------------------- *voxox.com <http://www.voxox.com/> * 5825 Oberlin Drive, Suite 5 San Diego, CA 92121 [image: Voxox]
On Wed, Aug 8, 2018 at 10:43 AM, Carlos Alvarez <caalvarez at gmail.com> wrote:
Do most of you have the phones authenticate incoming calls? We haven't been, but occasionally find a router that has unfiltered full cone NAT (Cisco) or that puts one phone on 5060 with no filtering by IP. The result is that the phone will start ringing at random as script kiddies hit the IP and port 5060 trying to find servers to exploit. I don't see a downside to changing to auth, but not having done it outside of a few tests of a small number of phones, I figured I would ask.
_______________________________________________ VoiceOps mailing list VoiceOps at voiceops.org https://puck.nether.net/mailman/listinfo/voiceops
-- Alex
-- Sent via mobile, please forgive typos and brevity. _______________________________________________ VoiceOps mailing list VoiceOps at voiceops.org https://puck.nether.net/mailman/listinfo/voiceops
-- Alex
-- Sent via mobile, please forgive typos and brevity. _______________________________________________ VoiceOps mailing list VoiceOps at voiceops.org https://puck.nether.net/mailman/listinfo/voiceops
TCP is definitely the way to go nowadays. We use TCP on Grandstreams all the time, especially on their ATAs. Speaking of which, switching from UDP to TCP will reduce your customers' support calls dramatically. I don't know the current status over there, but 2 years ago RingCentral moved to TCP as well: https://netstorage.ringcentral.com/documents/sip.pdf Aviv On Wed, Aug 8, 2018, at 12:21 PM, Carlos Alvarez wrote:
It has, but it wasn't that long ago that people were still having challenges. Our preferred phone vendor, Grandstream, still generally advises against it.> So...who else on the list uses TCP and has any comments about it?
On Wed, Aug 8, 2018 at 11:12 AM Alex Balashov <abalashov at evaristesys.com> wrote:>> That has changed greatly since 2005.
On August 8, 2018 2:07:50 PM EDT, Carlos Alvarez <caalvarez at gmail.com> wrote:>> >That's a change I've never investigated. Or more precisely,
haven't>> >investigated since the days when the advice for doing it was "good>> >luck!!"
On Wed, Aug 8, 2018 at 11:00 AM Alex Balashov <abalashov at evaristesys.com> wrote:
I would have to agree with Calvin. Just use TCP.
On August 8, 2018 1:58:47 PM EDT, Calvin Ellison <calvin.ellison at voxox.com> wrote:
Using TCP or TLS would avoid open NAT issue, and can cure some naughty SIP ALG issues as well, assuming you want to tolerate the overhead.>> >> > For UDP, we've used both Digest and Source request validation with>> >> >Polycom devices. Source validation is probably the easiest route, assuming>> >the UA doesn't need to receive calls from anyone but its proxy or registrar. Digest (nonce challenge) is better if you want to accept calls from>> >> >anyone who knows your password, but we had an issue with a softswitch that>> >> >would properly handle auth channel to INVITE but choked when a BYE was>> >> >challenged.
Regards,
*Calvin Ellison* Voice Operations Engineer calvin.ellison at voxox.com +1 (213) 285-0555
----------------------------------------------- *voxox.com <http://www.voxox.com/> * 5825 Oberlin Drive, Suite 5 San Diego, CA 92121 [image: Voxox]
On Wed, Aug 8, 2018 at 10:43 AM, Carlos Alvarez <caalvarez at gmail.com> wrote:
Do most of you have the phones authenticate incoming calls? We>> >> >haven't been, but occasionally find a router that has unfiltered full cone>> >> >NAT (Cisco) or that puts one phone on 5060 with no filtering by IP.>> >The result is that the phone will start ringing at random as script kiddies>> >hit the IP and port 5060 trying to find servers to exploit. I don't see a>> >> >downside to changing to auth, but not having done it outside of a few tests of>> >a small number of phones, I figured I would ask.
_______________________________________________ VoiceOps mailing list VoiceOps at voiceops.org https://puck.nether.net/mailman/listinfo/voiceops
-- Alex
-- Sent via mobile, please forgive typos and brevity. _______________________________________________ VoiceOps mailing list VoiceOps at voiceops.org https://puck.nether.net/mailman/listinfo/voiceops
-- Alex
-- Sent via mobile, please forgive typos and brevity. _______________________________________________ VoiceOps mailing list VoiceOps at voiceops.org https://puck.nether.net/mailman/listinfo/voiceops
_________________________________________________ VoiceOps mailing list VoiceOps at voiceops.org https://puck.nether.net/mailman/listinfo/voiceops
On Wed, Aug 08, 2018 at 12:21:09PM -0700, Carlos Alvarez wrote:
So...who else on the list uses TCP and has any comments about it?
We are not an ITSP and are Polycom-only with a trivial number of endpoints, but we do use it and it works just fine. However, we have numerous customers, some of whom use TCP predominantly for thousands of endpoints. It works just fine. In terms of downsides: In addition to a historical lack of (RFC 3261-mandated) support, there are of course theoretical trade-offs involved in using TCP. There's more overhead, and connection state to be maintained on the server side, which of course consumes resources ? resources considered trivial nowadays, but once upon a time, when RFC 3261 was ratified (2002), perhaps not. As with all things TCP, it can also present a DoS vector if you don't limit the number of connections somewhere. The congestion control/end-to-end delay aspects of TCP are certainly not as important now as they were at a time when the public IP backbone and was in an entirely different place in its evolution. Also, nowadays the congestion/windowing algorithms used in TCP can be tweaked to something more efficient. I think the most damning thing about using TCP is perceived to be the relative difficulty of failing over TCP session state to a different host. UDP does not require connection state, so as long as you have some means of handling requests in a relatively stateless fashion, things can just carry on as they did before in the event of an IP takeover without anyone having to "reconnect". This is one area where the big enterprise boxes certainly trump the open-source ecosystem, where transparent TCP failover *for SIP* doesn't really exist, although in my opinion the whole issue is getting a bit moot with the way cloud infrastructure and virtualisation networking is evolving. -- Alex -- Alex Balashov | Principal | Evariste Systems LLC Tel: +1-706-510-6800 / +1-800-250-5920 (toll-free) Web: http://www.evaristesys.com/, http://www.csrpswitch.com/
So those of you using TCP, are you also using TLS? On Wed, Aug 8, 2018 at 12:36 PM Alex Balashov <abalashov at evaristesys.com> wrote:
On Wed, Aug 08, 2018 at 12:21:09PM -0700, Carlos Alvarez wrote:
So...who else on the list uses TCP and has any comments about it?
We are not an ITSP and are Polycom-only with a trivial number of endpoints, but we do use it and it works just fine.
However, we have numerous customers, some of whom use TCP predominantly for thousands of endpoints. It works just fine.
In terms of downsides:
In addition to a historical lack of (RFC 3261-mandated) support, there are of course theoretical trade-offs involved in using TCP. There's more overhead, and connection state to be maintained on the server side, which of course consumes resources ? resources considered trivial nowadays, but once upon a time, when RFC 3261 was ratified (2002), perhaps not. As with all things TCP, it can also present a DoS vector if you don't limit the number of connections somewhere.
The congestion control/end-to-end delay aspects of TCP are certainly not as important now as they were at a time when the public IP backbone and was in an entirely different place in its evolution. Also, nowadays the congestion/windowing algorithms used in TCP can be tweaked to something more efficient.
I think the most damning thing about using TCP is perceived to be the relative difficulty of failing over TCP session state to a different host. UDP does not require connection state, so as long as you have some means of handling requests in a relatively stateless fashion, things can just carry on as they did before in the event of an IP takeover without anyone having to "reconnect". This is one area where the big enterprise boxes certainly trump the open-source ecosystem, where transparent TCP failover *for SIP* doesn't really exist, although in my opinion the whole issue is getting a bit moot with the way cloud infrastructure and virtualisation networking is evolving.
-- Alex
-- Alex Balashov | Principal | Evariste Systems LLC
Tel: +1-706-510-6800 / +1-800-250-5920 (toll-free) Web: http://www.evaristesys.com/, http://www.csrpswitch.com/
On Wed, Aug 08, 2018 at 12:39:17PM -0700, Carlos Alvarez wrote:
So those of you using TCP, are you also using TLS?
For some customers who require it, it's done, though as we both know, it's silly since you can only provide encryption on the last mile. But no, SIP-TLS is a whole different ball of wax with an entirely different reliability, interoperability and practicality profile, and one I would not advise you to get into unnecessarily. -- Alex Balashov | Principal | Evariste Systems LLC Tel: +1-706-510-6800 / +1-800-250-5920 (toll-free) Web: http://www.evaristesys.com/, http://www.csrpswitch.com/
OK so to expand on my previous smug-ness Upsides: * No more signaling nat issues. Literally zero. If you want to be super-sneaky run your edge over TLS port 443 and most things wont touch you. * No retransmission's or registration avalanches. They simply cannot happen since you need a tcp session first. * No packet fragmentation issues. Send massive bloated SDP's and never worry about pruning headers again. If you are doing sip SIMPLE send mime bodies in messages if you want. Its all good. * Faster convergence (if you reset the TCP connections to your devices it will usually trigger an instantaneous proxy advance) * Real-HA on carrier grade SBC's works just fine and TCP state is maintained across pairs (Acme, Perimeta etc) * Never chase lost signaling Downsides: * Conventional HA doesnt work so well. Your reg/subscription etc will all be in the context of a single TCP session (with or without TLS). This means for that second when you restart your proxy the session is lost and MUST be re-establised by the client. * SIP Outbound support, which would basically be the answer here basically doesn't exist in a usable fashion for reliable dual-reg. Device support is partial and broken. Its not good. There are potential solutions but it involves real commitment to this right now and the gulf of experience between having and not isnt huge. * Moderately more load since TCP state must be retained, but on modern hardware this is so trivial its almost not worth mentioning. * Need to re-learn KPI's for network. The entire signaling profile changes. Its just a different animal. * Most of your sniffer-based diagnostic tools become useless (for tls) since packets wont be readable. This is dodged with an edge that will feed encrypted traffic to a collector. Suggestions: STRONGLY recommend terminating TCP/TLS at the edge and still running core in straight UDP with jumbo frames. You dont want a cascde of tcp session reestablishments I have a growing SP network today doing this with great success and also advise my consulting clients to take this path. On 8/8/2018 12:36 PM, Alex Balashov wrote:
On Wed, Aug 08, 2018 at 12:21:09PM -0700, Carlos Alvarez wrote:
So...who else on the list uses TCP and has any comments about it? We are not an ITSP and are Polycom-only with a trivial number of endpoints, but we do use it and it works just fine.
However, we have numerous customers, some of whom use TCP predominantly for thousands of endpoints. It works just fine.
In terms of downsides:
In addition to a historical lack of (RFC 3261-mandated) support, there are of course theoretical trade-offs involved in using TCP. There's more overhead, and connection state to be maintained on the server side, which of course consumes resources ? resources considered trivial nowadays, but once upon a time, when RFC 3261 was ratified (2002), perhaps not. As with all things TCP, it can also present a DoS vector if you don't limit the number of connections somewhere.
The congestion control/end-to-end delay aspects of TCP are certainly not as important now as they were at a time when the public IP backbone and was in an entirely different place in its evolution. Also, nowadays the congestion/windowing algorithms used in TCP can be tweaked to something more efficient.
I think the most damning thing about using TCP is perceived to be the relative difficulty of failing over TCP session state to a different host. UDP does not require connection state, so as long as you have some means of handling requests in a relatively stateless fashion, things can just carry on as they did before in the event of an IP takeover without anyone having to "reconnect". This is one area where the big enterprise boxes certainly trump the open-source ecosystem, where transparent TCP failover *for SIP* doesn't really exist, although in my opinion the whole issue is getting a bit moot with the way cloud infrastructure and virtualisation networking is evolving.
-- Alex
Agree with everything Ryan said, with the caveat that TLS for TLS's sake is, in my own humble opinion, a terrible idea from a troubleshooting and general complexity perspective. Use where absolutely necessary and nowhere else. On August 8, 2018 7:37:13 PM EDT, Ryan Delgrosso <ryandelgrosso at gmail.com> wrote:
OK so to expand on my previous smug-ness
Upsides:
* No more signaling nat issues. Literally zero. If you want to be super-sneaky run your edge over TLS port 443 and most things wont touch you. * No retransmission's or registration avalanches. They simply cannot happen since you need a tcp session first. * No packet fragmentation issues. Send massive bloated SDP's and never worry about pruning headers again. If you are doing sip SIMPLE send mime bodies in messages if you want. Its all good. * Faster convergence (if you reset the TCP connections to your devices it will usually trigger an instantaneous proxy advance) * Real-HA on carrier grade SBC's works just fine and TCP state is maintained across pairs (Acme, Perimeta etc) * Never chase lost signaling
Downsides:
* Conventional HA doesnt work so well. Your reg/subscription etc will all be in the context of a single TCP session (with or without TLS). This means for that second when you restart your proxy the session is lost and MUST be re-establised by the client. * SIP Outbound support, which would basically be the answer here basically doesn't exist in a usable fashion for reliable dual-reg. Device support is partial and broken. Its not good. There are potential solutions but it involves real commitment to this right now and the gulf of experience between having and not isnt huge. * Moderately more load since TCP state must be retained, but on modern hardware this is so trivial its almost not worth mentioning. * Need to re-learn KPI's for network. The entire signaling profile changes. Its just a different animal. * Most of your sniffer-based diagnostic tools become useless (for tls) since packets wont be readable. This is dodged with an edge that will feed encrypted traffic to a collector.
Suggestions:
STRONGLY recommend terminating TCP/TLS at the edge and still running core in straight UDP with jumbo frames. You dont want a cascde of tcp session reestablishments
I have a growing SP network today doing this with great success and also advise my consulting clients to take this path.
On 8/8/2018 12:36 PM, Alex Balashov wrote:
On Wed, Aug 08, 2018 at 12:21:09PM -0700, Carlos Alvarez wrote:
So...who else on the list uses TCP and has any comments about it? We are not an ITSP and are Polycom-only with a trivial number of endpoints, but we do use it and it works just fine.
However, we have numerous customers, some of whom use TCP predominantly for thousands of endpoints. It works just fine.
In terms of downsides:
In addition to a historical lack of (RFC 3261-mandated) support, there are of course theoretical trade-offs involved in using TCP. There's more overhead, and connection state to be maintained on the server side, which of course consumes resources ? resources considered trivial nowadays, but once upon a time, when RFC 3261 was ratified (2002), perhaps not. As with all things TCP, it can also present a DoS vector if you don't limit the number of connections somewhere.
The congestion control/end-to-end delay aspects of TCP are certainly not as important now as they were at a time when the public IP backbone and was in an entirely different place in its evolution. Also, nowadays the congestion/windowing algorithms used in TCP can be tweaked to something more efficient.
I think the most damning thing about using TCP is perceived to be the relative difficulty of failing over TCP session state to a different host. UDP does not require connection state, so as long as you have some means of handling requests in a relatively stateless fashion, things can just carry on as they did before in the event of an IP takeover without anyone having to "reconnect". This is one area where the big enterprise boxes certainly trump the open-source ecosystem, where transparent TCP failover *for SIP* doesn't really exist, although in my opinion the whole issue is getting a bit moot with the way cloud infrastructure and virtualisation networking is evolving.
-- Alex
-- Alex -- Sent via mobile, please forgive typos and brevity.
I used to follow the same logic but recently have shifted. I now wholeheartedly follow the encrypt everywhere stance. Too many industries have compliance regulations where VoIP got the exemption because of grandfathered PSTN focused laws, but just because you CAN go plaintext doesnt make it the best answer, and its always stronger to say "yes" to the encryption question than "No but..." On 8/8/2018 5:14 PM, Alex Balashov wrote:
Agree with everything Ryan said, with the caveat that TLS for TLS's sake is, in my own humble opinion, a terrible idea from a troubleshooting and general complexity perspective. Use where absolutely necessary and nowhere else.
On August 8, 2018 7:37:13 PM EDT, Ryan Delgrosso <ryandelgrosso at gmail.com> wrote:
OK so to expand on my previous smug-ness
Upsides:
* No more signaling nat issues. Literally zero. If you want to be super-sneaky run your edge over TLS port 443 and most things wont touch you. * No retransmission's or registration avalanches. They simply cannot happen since you need a tcp session first. * No packet fragmentation issues. Send massive bloated SDP's and never worry about pruning headers again. If you are doing sip SIMPLE send mime bodies in messages if you want. Its all good. * Faster convergence (if you reset the TCP connections to your devices it will usually trigger an instantaneous proxy advance) * Real-HA on carrier grade SBC's works just fine and TCP state is maintained across pairs (Acme, Perimeta etc) * Never chase lost signaling
Downsides:
* Conventional HA doesnt work so well. Your reg/subscription etc will all be in the context of a single TCP session (with or without TLS). This means for that second when you restart your proxy the session is lost and MUST be re-establised by the client. * SIP Outbound support, which would basically be the answer here basically doesn't exist in a usable fashion for reliable dual-reg. Device support is partial and broken. Its not good. There are potential solutions but it involves real commitment to this right now and the gulf of experience between having and not isnt huge. * Moderately more load since TCP state must be retained, but on modern hardware this is so trivial its almost not worth mentioning. * Need to re-learn KPI's for network. The entire signaling profile changes. Its just a different animal. * Most of your sniffer-based diagnostic tools become useless (for tls) since packets wont be readable. This is dodged with an edge that will feed encrypted traffic to a collector.
Suggestions:
STRONGLY recommend terminating TCP/TLS at the edge and still running core in straight UDP with jumbo frames. You dont want a cascde of tcp session reestablishments
I have a growing SP network today doing this with great success and also advise my consulting clients to take this path.
On 8/8/2018 12:36 PM, Alex Balashov wrote:
On Wed, Aug 08, 2018 at 12:21:09PM -0700, Carlos Alvarez wrote:
So...who else on the list uses TCP and has any comments about it? We are not an ITSP and are Polycom-only with a trivial number of endpoints, but we do use it and it works just fine.
However, we have numerous customers, some of whom use TCP predominantly for thousands of endpoints. It works just fine.
In terms of downsides:
In addition to a historical lack of (RFC 3261-mandated) support, there are of course theoretical trade-offs involved in using TCP. There's more overhead, and connection state to be maintained on the server side, which of course consumes resources ? resources considered trivial nowadays, but once upon a time, when RFC 3261 was ratified (2002), perhaps not. As with all things TCP, it can also present a DoS vector if you don't limit the number of connections somewhere.
The congestion control/end-to-end delay aspects of TCP are certainly not as important now as they were at a time when the public IP backbone and was in an entirely different place in its evolution. Also, nowadays the congestion/windowing algorithms used in TCP can be tweaked to something more efficient.
I think the most damning thing about using TCP is perceived to be the relative difficulty of failing over TCP session state to a different host. UDP does not require connection state, so as long as you have some means of handling requests in a relatively stateless fashion, things can just carry on as they did before in the event of an IP takeover without anyone having to "reconnect". This is one area where the big enterprise boxes certainly trump the open-source ecosystem, where transparent TCP failover *for SIP* doesn't really exist, although in my opinion the whole issue is getting a bit moot with the way cloud infrastructure and virtualisation networking is evolving.
-- Alex
-- Alex
-- Sent via mobile, please forgive typos and brevity. _______________________________________________ VoiceOps mailing list VoiceOps at voiceops.org https://puck.nether.net/mailman/listinfo/voiceops
Yes, but until and unless your upstream supply chain is doing TLS and you can provide end-to-end security, it's a pointless waste of time. My customers have numerous customers who "require" "encryption" and "security", and this is provided to them on the "Last Mile" SIP trunk. But as soon as it goes to the usual Bandwidths and friends all that TLS is sheathed off. As long as that is the case, and I expect it will be the case for quite some time, this whole concept is a joke. The second problem is how incredibly inconsistent/broken SIP-TLS is. It's a trainwreck with way too many moving parts. My finding over the years has been that when it comes to providing faux-"security", my happiest customers are the ones that settled for a tunnel-based approach. -- Alex On Wed, Aug 08, 2018 at 10:09:40PM -0700, Ryan Delgrosso wrote:
I used to follow the same logic but recently have shifted. I now wholeheartedly follow the encrypt everywhere stance. Too many industries have compliance regulations where VoIP got the exemption because of grandfathered PSTN focused laws, but just because you CAN go plaintext doesnt make it the best answer, and its always stronger to say "yes" to the encryption question than "No but..."
On 8/8/2018 5:14 PM, Alex Balashov wrote:
Agree with everything Ryan said, with the caveat that TLS for TLS's sake is, in my own humble opinion, a terrible idea from a troubleshooting and general complexity perspective. Use where absolutely necessary and nowhere else.
On August 8, 2018 7:37:13 PM EDT, Ryan Delgrosso <ryandelgrosso at gmail.com> wrote:
OK so to expand on my previous smug-ness
Upsides:
* No more signaling nat issues. Literally zero. If you want to be super-sneaky run your edge over TLS port 443 and most things wont touch you. * No retransmission's or registration avalanches. They simply cannot happen since you need a tcp session first. * No packet fragmentation issues. Send massive bloated SDP's and never worry about pruning headers again. If you are doing sip SIMPLE send mime bodies in messages if you want. Its all good. * Faster convergence (if you reset the TCP connections to your devices it will usually trigger an instantaneous proxy advance) * Real-HA on carrier grade SBC's works just fine and TCP state is maintained across pairs (Acme, Perimeta etc) * Never chase lost signaling
Downsides:
* Conventional HA doesnt work so well. Your reg/subscription etc will all be in the context of a single TCP session (with or without TLS). This means for that second when you restart your proxy the session is lost and MUST be re-establised by the client. * SIP Outbound support, which would basically be the answer here basically doesn't exist in a usable fashion for reliable dual-reg. Device support is partial and broken. Its not good. There are potential solutions but it involves real commitment to this right now and the gulf of experience between having and not isnt huge. * Moderately more load since TCP state must be retained, but on modern hardware this is so trivial its almost not worth mentioning. * Need to re-learn KPI's for network. The entire signaling profile changes. Its just a different animal. * Most of your sniffer-based diagnostic tools become useless (for tls) since packets wont be readable. This is dodged with an edge that will feed encrypted traffic to a collector.
Suggestions:
STRONGLY recommend terminating TCP/TLS at the edge and still running core in straight UDP with jumbo frames. You dont want a cascde of tcp session reestablishments
I have a growing SP network today doing this with great success and also advise my consulting clients to take this path.
On 8/8/2018 12:36 PM, Alex Balashov wrote:
On Wed, Aug 08, 2018 at 12:21:09PM -0700, Carlos Alvarez wrote:
So...who else on the list uses TCP and has any comments about it? We are not an ITSP and are Polycom-only with a trivial number of endpoints, but we do use it and it works just fine.
However, we have numerous customers, some of whom use TCP predominantly for thousands of endpoints. It works just fine.
In terms of downsides:
In addition to a historical lack of (RFC 3261-mandated) support, there are of course theoretical trade-offs involved in using TCP. There's more overhead, and connection state to be maintained on the server side, which of course consumes resources ? resources considered trivial nowadays, but once upon a time, when RFC 3261 was ratified (2002), perhaps not. As with all things TCP, it can also present a DoS vector if you don't limit the number of connections somewhere.
The congestion control/end-to-end delay aspects of TCP are certainly not as important now as they were at a time when the public IP backbone and was in an entirely different place in its evolution. Also, nowadays the congestion/windowing algorithms used in TCP can be tweaked to something more efficient.
I think the most damning thing about using TCP is perceived to be the relative difficulty of failing over TCP session state to a different host. UDP does not require connection state, so as long as you have some means of handling requests in a relatively stateless fashion, things can just carry on as they did before in the event of an IP takeover without anyone having to "reconnect". This is one area where the big enterprise boxes certainly trump the open-source ecosystem, where transparent TCP failover *for SIP* doesn't really exist, although in my opinion the whole issue is getting a bit moot with the way cloud infrastructure and virtualisation networking is evolving.
-- Alex
-- Alex
-- Sent via mobile, please forgive typos and brevity. _______________________________________________ VoiceOps mailing list VoiceOps at voiceops.org https://puck.nether.net/mailman/listinfo/voiceops
_______________________________________________ VoiceOps mailing list VoiceOps at voiceops.org https://puck.nether.net/mailman/listinfo/voiceops
-- Alex Balashov | Principal | Evariste Systems LLC Tel: +1-706-510-6800 / +1-800-250-5920 (toll-free) Web: http://www.evaristesys.com/, http://www.csrpswitch.com/
Alex, what has been your experience with tunnel based solutions? Our choice seems to be IPSec VPN on existing gear or spending some cash on an SRTP-RTP "transcoder". Regards, *Calvin Ellison* Voice Operations Engineer calvin.ellison at voxox.com +1 (213) 285-0555 ----------------------------------------------- *voxox.com <http://www.voxox.com/> * 5825 Oberlin Drive, Suite 5 San Diego, CA 92121 [image: Voxox] On Thu, Aug 9, 2018 at 1:46 AM, Alex Balashov <abalashov at evaristesys.com> wrote:
Yes, but until and unless your upstream supply chain is doing TLS and you can provide end-to-end security, it's a pointless waste of time.
My customers have numerous customers who "require" "encryption" and "security", and this is provided to them on the "Last Mile" SIP trunk. But as soon as it goes to the usual Bandwidths and friends all that TLS is sheathed off.
As long as that is the case, and I expect it will be the case for quite some time, this whole concept is a joke.
The second problem is how incredibly inconsistent/broken SIP-TLS is. It's a trainwreck with way too many moving parts. My finding over the years has been that when it comes to providing faux-"security", my happiest customers are the ones that settled for a tunnel-based approach.
-- Alex
On Wed, Aug 08, 2018 at 10:09:40PM -0700, Ryan Delgrosso wrote:
I used to follow the same logic but recently have shifted. I now wholeheartedly follow the encrypt everywhere stance. Too many industries have compliance regulations where VoIP got the exemption because of grandfathered PSTN focused laws, but just because you CAN go plaintext doesnt make it the best answer, and its always stronger to say "yes" to the encryption question than "No but..."
On 8/8/2018 5:14 PM, Alex Balashov wrote:
Agree with everything Ryan said, with the caveat that TLS for TLS's sake is, in my own humble opinion, a terrible idea from a troubleshooting and general complexity perspective. Use where absolutely necessary and nowhere else.
On August 8, 2018 7:37:13 PM EDT, Ryan Delgrosso < ryandelgrosso at gmail.com> wrote:
OK so to expand on my previous smug-ness
Upsides:
* No more signaling nat issues. Literally zero. If you want to be super-sneaky run your edge over TLS port 443 and most things wont touch you. * No retransmission's or registration avalanches. They simply cannot happen since you need a tcp session first. * No packet fragmentation issues. Send massive bloated SDP's and never worry about pruning headers again. If you are doing sip SIMPLE send mime bodies in messages if you want. Its all good. * Faster convergence (if you reset the TCP connections to your devices it will usually trigger an instantaneous proxy advance) * Real-HA on carrier grade SBC's works just fine and TCP state is maintained across pairs (Acme, Perimeta etc) * Never chase lost signaling
Downsides:
* Conventional HA doesnt work so well. Your reg/subscription etc will all be in the context of a single TCP session (with or without TLS). This means for that second when you restart your proxy the session is lost and MUST be re-establised by the client. * SIP Outbound support, which would basically be the answer here basically doesn't exist in a usable fashion for reliable dual-reg. Device support is partial and broken. Its not good. There are potential solutions but it involves real commitment to this right now and the gulf of experience between having and not isnt huge. * Moderately more load since TCP state must be retained, but on modern hardware this is so trivial its almost not worth mentioning. * Need to re-learn KPI's for network. The entire signaling profile changes. Its just a different animal. * Most of your sniffer-based diagnostic tools become useless (for tls) since packets wont be readable. This is dodged with an edge that will feed encrypted traffic to a collector.
Suggestions:
STRONGLY recommend terminating TCP/TLS at the edge and still running core in straight UDP with jumbo frames. You dont want a cascde of tcp session reestablishments
I have a growing SP network today doing this with great success and also advise my consulting clients to take this path.
On 8/8/2018 12:36 PM, Alex Balashov wrote:
On Wed, Aug 08, 2018 at 12:21:09PM -0700, Carlos Alvarez wrote:
So...who else on the list uses TCP and has any comments about it? We are not an ITSP and are Polycom-only with a trivial number of endpoints, but we do use it and it works just fine.
However, we have numerous customers, some of whom use TCP predominantly for thousands of endpoints. It works just fine.
In terms of downsides:
In addition to a historical lack of (RFC 3261-mandated) support, there are of course theoretical trade-offs involved in using TCP. There's more overhead, and connection state to be maintained on the server side, which of course consumes resources ? resources considered trivial nowadays, but once upon a time, when RFC 3261 was ratified (2002), perhaps not. As with all things TCP, it can also present a DoS vector if you don't limit the number of connections somewhere.
The congestion control/end-to-end delay aspects of TCP are certainly not as important now as they were at a time when the public IP backbone and was in an entirely different place in its evolution. Also, nowadays the congestion/windowing algorithms used in TCP can be tweaked to something more efficient.
I think the most damning thing about using TCP is perceived to be the relative difficulty of failing over TCP session state to a different host. UDP does not require connection state, so as long as you have some means of handling requests in a relatively stateless fashion, things can just carry on as they did before in the event of an IP takeover without anyone having to "reconnect". This is one area where the big enterprise boxes certainly trump the open-source ecosystem, where transparent TCP failover *for SIP* doesn't really exist, although in my opinion the whole issue is getting a bit moot with the way cloud infrastructure and virtualisation networking is evolving.
-- Alex
-- Alex
-- Sent via mobile, please forgive typos and brevity. _______________________________________________ VoiceOps mailing list VoiceOps at voiceops.org https://puck.nether.net/mailman/listinfo/voiceops
_______________________________________________ VoiceOps mailing list VoiceOps at voiceops.org https://puck.nether.net/mailman/listinfo/voiceops
-- Alex Balashov | Principal | Evariste Systems LLC
Tel: +1-706-510-6800 / +1-800-250-5920 (toll-free) Web: http://www.evaristesys.com/, http://www.csrpswitch.com/ _______________________________________________ VoiceOps mailing list VoiceOps at voiceops.org https://puck.nether.net/mailman/listinfo/voiceops
On Thu, Aug 09, 2018 at 10:57:36AM -0700, Calvin Ellison wrote:
Alex, what has been your experience with tunnel based solutions? Our choice seems to be IPSec VPN on existing gear or spending some cash on an SRTP-RTP "transcoder".
I personally find IPSec way too complicated. My preference is to do OpenVPN, because it's so, so much simpler. There is at least one handset vendor - Snom - that supports it right in the handset. I've seen lots of tunnel-based approaches with hosted PBX. A lot of them involve sending the customer a router to put on their network where the tunnels can land and the traffic can be diverted into the tunnel. Others involve putting it straight into the phones, or in the case of softphones, packaging it with the UA. Whatever it is, it works better and is easier to set up and make behave more consistently than SIP-TLS. -- Alex Balashov | Principal | Evariste Systems LLC Tel: +1-706-510-6800 / +1-800-250-5920 (toll-free) Web: http://www.evaristesys.com/, http://www.csrpswitch.com/
On the security side...there's reality and there's perception. I go through HIPAA/PCI compliance docs all the time, and they ask things about encryption or secure/dedicated connections. I can say "no, but...." or I can just say "yes." Our HIPPA-covered clients are all on MPLS to us so we can answer yes. When I say "no but" on other topics, there's always a hassle. There is a practical matter that there are probably more threats on the last mile than anywhere else. Someone at Level 3 is unlikely to be sniffing on a large connection, but consumer ISP employees have much easier access to the data. Downstream from there, plenty of places where there is little security. So "secure to the first server" does have a certain value. Either way, I don't intend to use TLS. It was more of an academic question. On Thu, Aug 9, 2018 at 1:48 AM Alex Balashov <abalashov at evaristesys.com> wrote:
Yes, but until and unless your upstream supply chain is doing TLS and you can provide end-to-end security, it's a pointless waste of time.
My customers have numerous customers who "require" "encryption" and "security", and this is provided to them on the "Last Mile" SIP trunk. But as soon as it goes to the usual Bandwidths and friends all that TLS is sheathed off.
As long as that is the case, and I expect it will be the case for quite some time, this whole concept is a joke.
The second problem is how incredibly inconsistent/broken SIP-TLS is. It's a trainwreck with way too many moving parts. My finding over the years has been that when it comes to providing faux-"security", my happiest customers are the ones that settled for a tunnel-based approach.
-- Alex
On Wed, Aug 08, 2018 at 10:09:40PM -0700, Ryan Delgrosso wrote:
I used to follow the same logic but recently have shifted. I now wholeheartedly follow the encrypt everywhere stance. Too many industries have compliance regulations where VoIP got the exemption because of grandfathered PSTN focused laws, but just because you CAN go plaintext doesnt make it the best answer, and its always stronger to say "yes" to the encryption question than "No but..."
On 8/8/2018 5:14 PM, Alex Balashov wrote:
Agree with everything Ryan said, with the caveat that TLS for TLS's sake is, in my own humble opinion, a terrible idea from a troubleshooting and general complexity perspective. Use where absolutely necessary and nowhere else.
On August 8, 2018 7:37:13 PM EDT, Ryan Delgrosso < ryandelgrosso at gmail.com> wrote:
OK so to expand on my previous smug-ness
Upsides:
* No more signaling nat issues. Literally zero. If you want to be super-sneaky run your edge over TLS port 443 and most things wont touch you. * No retransmission's or registration avalanches. They simply cannot happen since you need a tcp session first. * No packet fragmentation issues. Send massive bloated SDP's and never worry about pruning headers again. If you are doing sip SIMPLE send mime bodies in messages if you want. Its all good. * Faster convergence (if you reset the TCP connections to your devices it will usually trigger an instantaneous proxy advance) * Real-HA on carrier grade SBC's works just fine and TCP state is maintained across pairs (Acme, Perimeta etc) * Never chase lost signaling
Downsides:
* Conventional HA doesnt work so well. Your reg/subscription etc will all be in the context of a single TCP session (with or without TLS). This means for that second when you restart your proxy the session is lost and MUST be re-establised by the client. * SIP Outbound support, which would basically be the answer here basically doesn't exist in a usable fashion for reliable dual-reg. Device support is partial and broken. Its not good. There are potential solutions but it involves real commitment to this right now and the gulf of experience between having and not isnt huge. * Moderately more load since TCP state must be retained, but on modern hardware this is so trivial its almost not worth mentioning. * Need to re-learn KPI's for network. The entire signaling profile changes. Its just a different animal. * Most of your sniffer-based diagnostic tools become useless (for tls) since packets wont be readable. This is dodged with an edge that will feed encrypted traffic to a collector.
Suggestions:
STRONGLY recommend terminating TCP/TLS at the edge and still running core in straight UDP with jumbo frames. You dont want a cascde of tcp session reestablishments
I have a growing SP network today doing this with great success and also advise my consulting clients to take this path.
On 8/8/2018 12:36 PM, Alex Balashov wrote:
On Wed, Aug 08, 2018 at 12:21:09PM -0700, Carlos Alvarez wrote:
So...who else on the list uses TCP and has any comments about it? We are not an ITSP and are Polycom-only with a trivial number of endpoints, but we do use it and it works just fine.
However, we have numerous customers, some of whom use TCP predominantly for thousands of endpoints. It works just fine.
In terms of downsides:
In addition to a historical lack of (RFC 3261-mandated) support, there are of course theoretical trade-offs involved in using TCP. There's more overhead, and connection state to be maintained on the server side, which of course consumes resources ? resources considered trivial nowadays, but once upon a time, when RFC 3261 was ratified (2002), perhaps not. As with all things TCP, it can also present a DoS vector if you don't limit the number of connections somewhere.
The congestion control/end-to-end delay aspects of TCP are certainly not as important now as they were at a time when the public IP backbone and was in an entirely different place in its evolution. Also, nowadays the congestion/windowing algorithms used in TCP can be tweaked to something more efficient.
I think the most damning thing about using TCP is perceived to be the relative difficulty of failing over TCP session state to a different host. UDP does not require connection state, so as long as you have some means of handling requests in a relatively stateless fashion, things can just carry on as they did before in the event of an IP takeover without anyone having to "reconnect". This is one area where the big enterprise boxes certainly trump the open-source ecosystem, where transparent TCP failover *for SIP* doesn't really exist, although in my opinion the whole issue is getting a bit moot with the way cloud infrastructure and virtualisation networking is evolving.
-- Alex
-- Alex
-- Sent via mobile, please forgive typos and brevity. _______________________________________________ VoiceOps mailing list VoiceOps at voiceops.org https://puck.nether.net/mailman/listinfo/voiceops
_______________________________________________ VoiceOps mailing list VoiceOps at voiceops.org https://puck.nether.net/mailman/listinfo/voiceops
-- Alex Balashov | Principal | Evariste Systems LLC
Tel: +1-706-510-6800 / +1-800-250-5920 (toll-free) Web: http://www.evaristesys.com/, http://www.csrpswitch.com/ _______________________________________________ VoiceOps mailing list VoiceOps at voiceops.org https://puck.nether.net/mailman/listinfo/voiceops
The "but look at them" argument never tasted good to me. Someone has to break rank and do it right first or it never gets better. There is also the point that intra-org communications can be end-to-end guarenteed. That resonates strongly. On 8/9/2018 1:46 AM, Alex Balashov wrote:
Yes, but until and unless your upstream supply chain is doing TLS and you can provide end-to-end security, it's a pointless waste of time.
My customers have numerous customers who "require" "encryption" and "security", and this is provided to them on the "Last Mile" SIP trunk. But as soon as it goes to the usual Bandwidths and friends all that TLS is sheathed off.
As long as that is the case, and I expect it will be the case for quite some time, this whole concept is a joke.
The second problem is how incredibly inconsistent/broken SIP-TLS is. It's a trainwreck with way too many moving parts. My finding over the years has been that when it comes to providing faux-"security", my happiest customers are the ones that settled for a tunnel-based approach.
-- Alex
On Wed, Aug 08, 2018 at 10:09:40PM -0700, Ryan Delgrosso wrote:
I used to follow the same logic but recently have shifted. I now wholeheartedly follow the encrypt everywhere stance. Too many industries have compliance regulations where VoIP got the exemption because of grandfathered PSTN focused laws, but just because you CAN go plaintext doesnt make it the best answer, and its always stronger to say "yes" to the encryption question than "No but..."
On 8/8/2018 5:14 PM, Alex Balashov wrote:
Agree with everything Ryan said, with the caveat that TLS for TLS's sake is, in my own humble opinion, a terrible idea from a troubleshooting and general complexity perspective. Use where absolutely necessary and nowhere else.
On August 8, 2018 7:37:13 PM EDT, Ryan Delgrosso <ryandelgrosso at gmail.com> wrote:
OK so to expand on my previous smug-ness
Upsides:
* No more signaling nat issues. Literally zero. If you want to be super-sneaky run your edge over TLS port 443 and most things wont touch you. * No retransmission's or registration avalanches. They simply cannot happen since you need a tcp session first. * No packet fragmentation issues. Send massive bloated SDP's and never worry about pruning headers again. If you are doing sip SIMPLE send mime bodies in messages if you want. Its all good. * Faster convergence (if you reset the TCP connections to your devices it will usually trigger an instantaneous proxy advance) * Real-HA on carrier grade SBC's works just fine and TCP state is maintained across pairs (Acme, Perimeta etc) * Never chase lost signaling
Downsides:
* Conventional HA doesnt work so well. Your reg/subscription etc will all be in the context of a single TCP session (with or without TLS). This means for that second when you restart your proxy the session is lost and MUST be re-establised by the client. * SIP Outbound support, which would basically be the answer here basically doesn't exist in a usable fashion for reliable dual-reg. Device support is partial and broken. Its not good. There are potential solutions but it involves real commitment to this right now and the gulf of experience between having and not isnt huge. * Moderately more load since TCP state must be retained, but on modern hardware this is so trivial its almost not worth mentioning. * Need to re-learn KPI's for network. The entire signaling profile changes. Its just a different animal. * Most of your sniffer-based diagnostic tools become useless (for tls) since packets wont be readable. This is dodged with an edge that will feed encrypted traffic to a collector.
Suggestions:
STRONGLY recommend terminating TCP/TLS at the edge and still running core in straight UDP with jumbo frames. You dont want a cascde of tcp session reestablishments
I have a growing SP network today doing this with great success and also advise my consulting clients to take this path.
On 8/8/2018 12:36 PM, Alex Balashov wrote:
On Wed, Aug 08, 2018 at 12:21:09PM -0700, Carlos Alvarez wrote:
So...who else on the list uses TCP and has any comments about it? We are not an ITSP and are Polycom-only with a trivial number of endpoints, but we do use it and it works just fine.
However, we have numerous customers, some of whom use TCP predominantly for thousands of endpoints. It works just fine.
In terms of downsides:
In addition to a historical lack of (RFC 3261-mandated) support, there are of course theoretical trade-offs involved in using TCP. There's more overhead, and connection state to be maintained on the server side, which of course consumes resources ? resources considered trivial nowadays, but once upon a time, when RFC 3261 was ratified (2002), perhaps not. As with all things TCP, it can also present a DoS vector if you don't limit the number of connections somewhere.
The congestion control/end-to-end delay aspects of TCP are certainly not as important now as they were at a time when the public IP backbone and was in an entirely different place in its evolution. Also, nowadays the congestion/windowing algorithms used in TCP can be tweaked to something more efficient.
I think the most damning thing about using TCP is perceived to be the relative difficulty of failing over TCP session state to a different host. UDP does not require connection state, so as long as you have some means of handling requests in a relatively stateless fashion, things can just carry on as they did before in the event of an IP takeover without anyone having to "reconnect". This is one area where the big enterprise boxes certainly trump the open-source ecosystem, where transparent TCP failover *for SIP* doesn't really exist, although in my opinion the whole issue is getting a bit moot with the way cloud infrastructure and virtualisation networking is evolving.
-- Alex
-- Alex
-- Sent via mobile, please forgive typos and brevity. _______________________________________________ VoiceOps mailing list VoiceOps at voiceops.org https://puck.nether.net/mailman/listinfo/voiceops
VoiceOps mailing list VoiceOps at voiceops.org https://puck.nether.net/mailman/listinfo/voiceops
On 08/09/2018 04:46 AM, Alex Balashov wrote:
Yes, but until and unless your upstream supply chain is doing TLS and you can provide end-to-end security, it's a pointless waste of time.
There's also an argument to be made that I haven't seen brought up for protecting SIP registration credentials either by providing transport confidentiality for a conventional password/secret or by using TLS client certificates. If you're at all worried about an adversary observing your actual comms, I'd be doubly worried about somebody stealing registration credentials and abusing them. -- Brandon Martin
On Aug 9, 2018, at 9:47 PM, Brandon Martin <lists.voiceops at monmotha.net> wrote:
On 08/09/2018 04:46 AM, Alex Balashov wrote:
Yes, but until and unless your upstream supply chain is doing TLS and you can provide end-to-end security, it's a pointless waste of time.
There's also an argument to be made that I haven't seen brought up for protecting SIP registration credentials either by providing transport confidentiality for a conventional password/secret or by using TLS client certificates. If you're at all worried about an adversary observing your actual comms, I'd be doubly worried about somebody stealing registration credentials and abusing them.
TLS was never about end to end confidentiality. We have wiretap obligations after all. Until the last copper line is dead and gone there will always be a way for unencrypted calls to occur. TLS is good when you don't want your local IT staff to know what the CEO is talking about, or to wiretap his coworkers (assuming hosted PBX). The likely attack surface for a customer's confidentiality will be somewhere between that handset and you, and you have a means to protect that. -Paul
I've seen similar issues with Polycom phones, the fix was to set voIpProt.SIP.strictUserValidation to "1". I don't use any other brands, so I don't know about others. On 08/08/2018 01:43 PM, Carlos Alvarez wrote:
Do most of you have the phones authenticate incoming calls?? We haven't been, but occasionally find a router that has unfiltered full cone NAT (Cisco) or that puts one phone on 5060 with no filtering by IP.? The result is that the phone will start ringing at random as script kiddies hit the IP and port 5060 trying to find servers to exploit.? I don't see a downside to changing to auth, but not having done it outside of a few tests of a small number of phones, I figured I would ask.
_______________________________________________ VoiceOps mailing list VoiceOps at voiceops.org https://puck.nether.net/mailman/listinfo/voiceops
Step 1: Move everything to TCP (or better yet TLS). Step 2: Use LetsEncrypt for free perpetual certificates. Step 3: Profit (and smug superiority over UDP based competitors battling nat issues). On 8/8/2018 10:43 AM, Carlos Alvarez wrote:
Do most of you have the phones authenticate incoming calls?? We haven't been, but occasionally find a router that has unfiltered full cone NAT (Cisco) or that puts one phone on 5060 with no filtering by IP.? The result is that the phone will start ringing at random as script kiddies hit the IP and port 5060 trying to find servers to exploit.? I don't see a downside to changing to auth, but not having done it outside of a few tests of a small number of phones, I figured I would ask.
_______________________________________________ VoiceOps mailing list VoiceOps at voiceops.org https://puck.nether.net/mailman/listinfo/voiceops
participants (8)
-
abalashov＠evaristesys.com -
aviv＠ironsip.com -
caalvarez＠gmail.com -
calvin.ellison＠voxox.com -
ewieling＠nyigc.com -
lists.voiceops＠monmotha.net -
paul＠timmins.net -
ryandelgrosso＠gmail.com