LTE: Why SGSN has interfaces with MME and SGW only not with PGW ?

Question

Instead of connection between SGSN and SGW why can't we have SGW with GGSN?

Answer 1

Hi,
This one is quite old, hope no one would mind me digging this up and answering.

Answer to this is very simple, but before giving a one liner there is a little bit of history behind it.
In 3g there were 2 main nodes SGSN and GGSN. GGSN was a camel carrying traffic and the SGSN was the brain... which was involved in carrying traffic.
Engs noticed it is not efficient, why the hell you want SGSN to carry the traffic? So they came up with Direct tunnel. Direct connection between GGSN and NodeB/Radio in general, to safe hops, processing power, delay(?).

This should already ring a bell.. Stop for a sec, think about it yourself, it will stay longer if you figure out answer on your own.

When the LTE was introduced there were 3 nodes, MME, SGW and PGW. PGW is camel carrying traffic, SGW is handicapped brain, and MME is The Brain.
There is no need in connecting MME with PGW. There is no thinking part in PGW, so connection with SGW is really enough.

Long story short, promised one-liner: because there is no need of doing so.

Answer 2

Well, As far I understood, the question is Why SGSN has to interface with MME & S-GW? SGSN is control plane mobility anchor for 3G/UMTS and in LTE it's counter part is MME. This scenario is true only in case of Direct Tunnel Function deployed. So for inter-network mobility we need both mobility anchors to exchange location/bearer info for user handover. However in case of legacy 3G deployment (without Direct Tunnel), SGSN has both Control Plane & Data Plane. So for Control Plane, SGSN has to interact with MME while for data plane, SGSN is required to interact with Data Plane mobility anchor in LTE architecture i.e. SGW.

SGSN doesn't connect with PGW.