SGSNs of 2G days brought their fair share of interesting problems to deal with. One of those complains you would regularly hear from your customers was “Your SGSN is not able to achieve the capacity you promised”
SGSN capacity planning was an extremely complex exercise, and involved extensive capacity benchmarking against hundreds of parameters by quality assurance teams. SGSNs of those days were designed to be the centre of both signalling and data-paths in 3gpp networks, and did ciphering too, thus reducing the overall SGSN capacity drastically. It was always difficult to estimate the right mix of different types of signalling events, packet-sizes and the ciphering algorithms used by UEs in real networks; and at some time or other, estimations failed and SGSNs in field ran out of capacity.
On top of this, operators mostly ended up having a higher per-user network cost due to presence to two nodes (SGSN and GGSN) handling data. That’s because, even as same data that passed through SGSN and then GGSN, operator had to buy capacity on both the nodes separately. This was especially true if an operator decided to buy SGSNs and GGSNs from different vendors.
And that’s how, for me, the need of CUPS was born. CUPS stands for “Control and User Plan Separation”, and has been touted as one of the fundamental design aspect of 5G networks. In this blog, I intend to summarise evolution of CUPS and the resulting packet core architectures across various 3GPP releases, and how I view the road ahead.
LinkedIn (to continue reading): https://www.linkedin.com/pulse/evolution-cups-3gpp-networks-varini-gupta/