At the end you said i hope it will be usefull ..... you cannot imagine how this amazing explanatory video was usefull to me ... very clear , straight to the point ... big big thanks for this effort sir .
Ive been supporting LTE for Cisco since its inception and trained many people on GTP concepts, but I will no longer do such explanations as yours trumps anything that I have done especially with the assistance of your slides, and so will be referring requests for knowledge in this area, including LTE call setups with your other videos, to your series of videos. Absolutely fantastic, and I dont give out compliments easily.
GTP-C UDP port: 2123 GTP-U UDP port: 2152 both gtpc and gtpu run on the s5 interface. In the control plane, the transmitting as well as the receiving side of both sgw and pgw have got to have the same IP address. In the user plane, the transmit side and recieving side (of both sgw and pgw) can have different IP addresses.
very useful sir...I request you to upload more videos in the future.......able to learn from u......very easy to understand....thanks a lot lot for sharing u r stuff.....
Hi Irfan, Thanks a lot for a detailed and neat explanation in all your videos...When are you uploading you next videos...Eagerly waiting for the new topics in lte or NR.
Thank you for your videos their are realy cool! One qouestion if the Mahmed want to setup several bearers for the each of them the new GTP-U tunnel will be created?
Hi Irfan, Thanks for this wonderful tutorial. One Question, Do gateways(PGW/SGW) also open socket ports for each tunnel Id or is it just a logical channel to segregate different users?
There is one UDP socket (per PGW-SGW) open and multiple TEIDs are send on that socket.Implementations may have multiple UDP sockets open for sending, with a different port number. But on each socket multiple TEIDs can be sent. The destination port number for GTP-U has to be 2152.
Thanks for this great tutorial. I have a small question though. Different users may be identified by their individual private IP addresses assigned by the P-GW. So, why exactly do we need TEID? Is it right to say that one IP session (with packets flowing between the SRC_IP and DEST_IP may have multiple GTP tunnels (for example there are one to one mapping between the EPS bearers and the corresponding GTP tunnels), that's why we need individual tunnel IDs?
Hi Vinit. Thanks for your comment. You are asking, imu "Is TEID *really* needed?" The answer is No, but this is a GTP design decision. Consider downlink between PGW and SGW and assume that there are two UEs in the system for which DL traffic needs to be sent from PGW to SGW, with IP address UE_IP1 and UE_IP2. PGW could have tunneled packet to SGW UDP destination tunnel 2152(outer IP: IP_PGW->IP_SGW,UDP_Port=2152), the SGW could look into the inner destination IP address and figured out which packet is for UE_1 and which one is for UE_2 and processed the packet accordingly. To enable this, the PGW would have to tell the SGW about each UE's IP address and SGW will need to have context indexed on UE's IP addresses and not TEIDs. However in GTP the user-plane tunnels are setup before each mobile is provided an IP address. Also, IP address is (typically) allocated by DHCP and can be removed on lease time-out. Given these considerations and some others, the GTP-U designers decided to use TEID that is not related to UE's IP address to determine which packet is for which user. One can design a tunneling protocol between PGW and SGW to be based on inner IP address (eg IP-in-IP tunneling or GRE tunneling), but that would require appropriate signaling to enable SGW to be aware of UE's IP addresses. The way GTP-U is designed, SGW (and eNB) are not expected to be aware of UE's IP addresses. Actually, in a GTPU tunnel (identified by TEID), you can put in any string of bytes, eg Ethernet frames or ATM or X.25 frames. :-).
Irfan Ali Thanks a lot for such a detailed explanation! This indeed, answered my question. I think the mechanism you explained is adopted when the EPS bearers terminate at the SGW, instead of the PGW. For example when the PMIP protocol is used over the S5/S8 interface. Please correct me if I'm wrong. :)
Irfan Ali Thanks a lot for such a detailed explanation! This indeed, answered my question. I think the mechanism you explained is adopted when the EPS bearers terminate at the SGW, instead of the PGW. For example when the PMIP protocol is used over the S5/S8 interface. Please correct me if I'm wrong. :)
Hi Vinit. Agree that with PMIP on S5/S8, EPS bearer is terminated in the SGW with a single GRE tunnel per UE between the SGW and PGW. There is still a *single* GRE tunnel *per UE* between SGW and PGW. (as opposed to possibly multiple GTP-U tunnels per UE, one for each EPS bearer of the UE for GTP-U). The TEIDs of GTP-U tunnel are replaced by GRE keys. There is one DL GRE key for DL tunnel (PGW->SGW) and a different GRE key for UL GRE tunnel (SGW->PGW).
If you use IP address for forwarding, you need routing tables. Then routing logic. Every time a new user comes up all tables in the circuit of nodes need updation. You need a suitable routing control plane, Thats lot of additional software.
Hi Irfan, Thanks for very informative tutorial, I have one query if UE has created all 11 bearers (max bearers limit per UE is 11) then Does it mean, there will be 11 GTPU tunnels between SGW-PGW, and same for eNODEB-SGW?
Hi Irfan, Couple of questions. 1. Are DRBs and dedicated bearer the same. 2. What is the use of drb 4,5,6,7,8 esta blishment in LTE. 3. Is it like more DRBs are activated based on throughput requirement ie. the more data ul/dl is ongoing more DRBs will be activated? Thanks in advance!
It may help to look at 40:00 of LTE attach part 2: th-cam.com/video/xWzajElyC5Q/w-d-xo.html 1. An EPS bearer is e2e (from UE to PGW) consists of three parts: (a) Data Radio bearer (DRB) from UE to eNB, (b) S1-U bearer between eNB and SGW, and (c) S5-U bearer between SGW and PGW. 2&3. There are two main reasons "bearers" are used in LTE. (a) Enabling UE to get access to two different "data domains". The two data-domains in LTE are "internet" and "ims" (ims for voice VoLTE support). The UE is provided two different IP@, IP@1 for internet and IP@2 for IMS. Two different "EPS bearers", with separate DRB, S1-U GTP tunnel and S5-U GTP tunnel are created to handle traffic for these two IP@. In the uplink (from UE), the UE filters IP traffic from IP@1 to EPS-bearer-1 and IP@2 to EPS-bearer-2. In the downlink the PGW filters traffic to destination IP@1 to EPS-bearer-1 and to IP@2 to EPS-bearer-2. (b) For providing different QoS to different traffic within the same data-domain, eg IMS. For IMS (VoLTE), the SIP-signaling traffic needs different QoS from the voice-media traffic. So two different "EPS bearers" are created between UE and the PGW, one for voice and one for SIP-signaling. As usual each bearer consists of DRB, S1-U and S5-U bearer. Hence, in the example we are using so far, UE communicating to Internet and IMS, the UE will have three bearers, (i) one for Internet IP@1, and two for IMS IP@2, (ii) IP@2-voice and (iii) IP@2-SIP signaling. For (ii) and (iii), the UE does filtering based on UDP/TCP port numbers in addition to IP@. Filtering is done on IP-5 tupple (source/destination IP@, source/destination port numbers and protocol-type) in the UE for uplink and more detailed DPI (that can go beyond the 5-tuple) in PGW for downlink. Hence, the setting up of multiple EPS bearers, is not based on throughput requirements, but IP@ and QoS requirements.
Hi Gideon: Without more information, not sure I can help all that much. There are some lists on linkedin and also from different training organizations where I have seen people ask deployment level questions. Unfortunately, I have not seen a stackoverflow kind of list for LTE, though there have been attempts to start one on stackexchange.
Dear Irfan Ail, It is very informative session that you have presented to us, also I noticed you did plan to release S1 Primer however unable to find one. Please do upload videos on 5G we are awaiting the same. Thank you for sharing the knowledge.
At the end you said i hope it will be usefull ..... you cannot imagine how this amazing explanatory video was usefull to me ... very clear , straight to the point ... big big thanks for this effort sir .
Best video available on TH-cam to understand gtp protocol
Ive been supporting LTE for Cisco since its inception and trained many people on GTP concepts, but I will no longer do such explanations as yours trumps anything that I have done especially with the assistance of your slides, and so will be referring requests for knowledge in this area, including LTE call setups with your other videos, to your series of videos. Absolutely fantastic, and I dont give out compliments easily.
I got a good understanding of the GTP concepts from this video. Thank you very much Irfan.
Great Video Sir
omggg the best explanation i've seen on yt
thank you Mr..Irfan for everything
Hello Sir, Excellent session, I forgot myself for 42min.
Good explanation, thanks Ali
GTP-C UDP port: 2123
GTP-U UDP port: 2152
both gtpc and gtpu run on the s5 interface.
In the control plane, the transmitting as well as the receiving side of both sgw and pgw have got to have the same IP address.
In the user plane, the transmit side and recieving side (of both sgw and pgw) can have different IP addresses.
Thanks !! Explained Gtp c and Gtp u very well!!
Really nice explanation. Thanks for publishing it!
You are welcome. Glad it was useful to you.
Looking forward for S1AP Primer
Irfan, there seems to be a typo around 20:47. Instead of the user plane TEID, the control plane TEID is being mentioned. Please check.
Excellent session, this is an eye opener for me! Continue your great work.
I desperately need 'S1AP primer'. Please. Please.
very useful sir...I request you to upload more videos in the future.......able to learn from u......very easy to understand....thanks a lot lot for sharing u r stuff.....
Wow, so simply explained. Thanks a lot for this video. Please let me know if there are more on this topic. or on GTP Prime.
Very useful. Thanks Irfan.
Very clear explanation. Thanks.
Wonderfull Irfan , thanks so much , u cleared me many doubts with your videos ;-)
Really a wonderful tutorial. Great job and appreciare your effort.
Very nice explanation.Thank you sir.
Hi Irfan,
Thanks a lot for a detailed and neat explanation in all your videos...When are you uploading you next videos...Eagerly waiting for the new topics in lte or NR.
Hi sir you are working on networking
Is it????
Thank you for your videos their are realy cool! One qouestion if the Mahmed want to setup several bearers for the each of them the new GTP-U tunnel will be created?
Sergey, yes there is one GTP-U tunnel for each bearer in EPC.
Hi Irfan, Thanks for this wonderful tutorial. One Question, Do gateways(PGW/SGW) also open socket ports for each tunnel Id or is it just a logical channel to segregate different users?
There is one UDP socket (per PGW-SGW) open and multiple TEIDs are send on that socket.Implementations may have multiple UDP sockets open for sending, with a different port number. But on each socket multiple TEIDs can be sent. The destination port number for GTP-U has to be 2152.
Thanks for a very precise video. Just a request... Can you have something similar for 5G
good explaination...
Thanks for this great tutorial. I have a small question though. Different users may be identified by their individual private IP addresses assigned by the P-GW. So, why exactly do we need TEID? Is it right to say that one IP session (with packets flowing between the SRC_IP and DEST_IP may have multiple GTP tunnels (for example there are one to one mapping between the EPS bearers and the corresponding GTP tunnels), that's why we need individual tunnel IDs?
Hi Vinit. Thanks for your comment. You are asking, imu "Is TEID *really* needed?" The answer is No, but this is a GTP design decision. Consider downlink between PGW and SGW and assume that there are two UEs in the system for which DL traffic needs to be sent from PGW to SGW, with IP address UE_IP1 and UE_IP2. PGW could have tunneled packet to SGW UDP destination tunnel 2152(outer IP: IP_PGW->IP_SGW,UDP_Port=2152), the SGW could look into the inner destination IP address and figured out which packet is for UE_1 and which one is for UE_2 and processed the packet accordingly. To enable this, the PGW would have to tell the SGW about each UE's IP address and SGW will need to have context indexed on UE's IP addresses and not TEIDs. However in GTP the user-plane tunnels are setup before each mobile is provided an IP address. Also, IP address is (typically) allocated by DHCP and can be removed on lease time-out. Given these considerations and some others, the GTP-U designers decided to use TEID that is not related to UE's IP address to determine which packet is for which user. One can design a tunneling protocol between PGW and SGW to be based on inner IP address (eg IP-in-IP tunneling or GRE tunneling), but that would require appropriate signaling to enable SGW to be aware of UE's IP addresses. The way GTP-U is designed, SGW (and eNB) are not expected to be aware of UE's IP addresses. Actually, in a GTPU tunnel (identified by TEID), you can put in any string of bytes, eg Ethernet frames or ATM or X.25 frames. :-).
Irfan Ali Thanks a lot for such a detailed explanation! This indeed, answered my question. I think the mechanism you explained is adopted when the EPS bearers terminate at the SGW, instead of the PGW. For example when the PMIP protocol is used over the S5/S8 interface. Please correct me if I'm wrong. :)
Irfan Ali Thanks a lot for such a detailed explanation! This indeed, answered my question. I think the mechanism you explained is adopted when the EPS bearers terminate at the SGW, instead of the PGW. For example when the PMIP protocol is used over the S5/S8 interface. Please correct me if I'm wrong. :)
Hi Vinit. Agree that with PMIP on S5/S8, EPS bearer is terminated in the SGW with a single GRE tunnel per UE between the SGW and PGW. There is still a *single* GRE tunnel *per UE* between SGW and PGW. (as opposed to possibly multiple GTP-U tunnels per UE, one for each EPS bearer of the UE for GTP-U). The TEIDs of GTP-U tunnel are replaced by GRE keys. There is one DL GRE key for DL tunnel (PGW->SGW) and a different GRE key for UL GRE tunnel (SGW->PGW).
If you use IP address for forwarding, you need routing tables. Then routing logic. Every time a new user comes up all tables in the circuit of nodes need updation. You need a suitable routing control plane, Thats lot of additional software.
Hi Irfan, First of all, many thanks for making such an amazing set of informational videos. Are you also going to post any videos relating to QOS ?
Yes I would love the QoS videos please? Fantastic explanations thanks!
excellent, please upload on VoLTE as well
Hi. You did a fantastic job. However, I desperately need 'S1AP primer'. Please upload 'S1AP primer' to this channel. Thank you.
Hi Irfan, Thanks for very informative tutorial, I have one query if UE has created all 11 bearers (max bearers limit per UE is 11) then Does it mean, there will be 11 GTPU tunnels between SGW-PGW, and same for eNODEB-SGW?
Yes.
Hi Irfan,
Couple of questions.
1. Are DRBs and dedicated bearer the same.
2. What is the use of drb 4,5,6,7,8 esta blishment in LTE.
3. Is it like more DRBs are activated based on throughput requirement ie. the more data ul/dl is ongoing more DRBs will be activated?
Thanks in advance!
It may help to look at 40:00 of LTE attach part 2: th-cam.com/video/xWzajElyC5Q/w-d-xo.html
1. An EPS bearer is e2e (from UE to PGW) consists of three parts: (a) Data Radio bearer (DRB) from UE to eNB, (b) S1-U bearer between eNB and SGW, and (c) S5-U bearer between SGW and PGW.
2&3. There are two main reasons "bearers" are used in LTE. (a) Enabling UE to get access to two different "data domains". The two data-domains in LTE are "internet" and "ims" (ims for voice VoLTE support). The UE is provided two different IP@, IP@1 for internet and IP@2 for IMS. Two different "EPS bearers", with separate DRB, S1-U GTP tunnel and S5-U GTP tunnel are created to handle traffic for these two IP@. In the uplink (from UE), the UE filters IP traffic from IP@1 to EPS-bearer-1 and IP@2 to EPS-bearer-2. In the downlink the PGW filters traffic to destination IP@1 to EPS-bearer-1 and to IP@2 to EPS-bearer-2.
(b) For providing different QoS to different traffic within the same data-domain, eg IMS. For IMS (VoLTE), the SIP-signaling traffic needs different QoS from the voice-media traffic. So two different "EPS bearers" are created between UE and the PGW, one for voice and one for SIP-signaling. As usual each bearer consists of DRB, S1-U and S5-U bearer. Hence, in the example we are using so far, UE communicating to Internet and IMS, the UE will have three bearers, (i) one for Internet IP@1, and two for IMS IP@2, (ii) IP@2-voice and (iii) IP@2-SIP signaling. For (ii) and (iii), the UE does filtering based on UDP/TCP port numbers in addition to IP@. Filtering is done on IP-5 tupple (source/destination IP@, source/destination port numbers and protocol-type) in the UE for uplink and more detailed DPI (that can go beyond the 5-tuple) in PGW for downlink.
Hence, the setting up of multiple EPS bearers, is not based on throughput requirements, but IP@ and QoS requirements.
Thanks a lot for the detailed answer 👍
Thanks Irfan Ali, this is quite an eye opener,
I have a issue with poor e-RAB success rate. which are the main pointers of cause?
Hi Gideon: Without more information, not sure I can help all that much. There are some lists on linkedin and also from different training organizations where I have seen people ask deployment level questions. Unfortunately, I have not seen a stackoverflow kind of list for LTE, though there have been attempts to start one on stackexchange.
thank you, sir. good work. really very good explanation.
Glad that you liked it :-)
Very Nice Tutorial. Can you please share same for Diameter protocol
Dear Irfan Ail,
It is very informative session that you have presented to us, also I noticed you did plan to release S1 Primer however unable to find one. Please do upload videos on 5G we are awaiting the same.
Thank you for sharing the knowledge.
Sorry have not created S1 primer. The attach video part 2 does provide some information on the connection IDs used on S1AP interface
This is great content. Any chance of having 5G contents :-)
Thanks a lot ...
excellent