As Mobile Network Operators (MNO) light up LTE services, they are witnessing a massive expansion in several dimensions, especially lured by the better quality of experience and increasing number of subscribers upgrading to LTE. In order to cater to the increasing influx of subscriber base, operators are adding higher bandwidth to their capacity, including Gi-. Newer LTE handsets are made with larger screen size, multi-core CPUs, higher capacity batteries and higher pixel density, features that offers greater user experience. Several newer mobile-only applications are taking advantage of this infrastructure, providing more value to subscribers. This has resulted in more subscribers switching to LTE and this is a cycle that will continue for a long time.
Traditionally operators built networks with limited visibility in place. Often the DPI box was a major component of visibility infrastructure. DPIs are typically deployed at Gi LAN, which offer great usage analysis. Operators also deploy 'Probes' and 'Tools' that decode various interfaces of a network, including core and RAN interfaces.
Probes are often a purpose-built hardware with rated capacity. As the mobile network operator makes incremental investments in Network Infrastructure such as eNodeB, MME, SPGW, and the capacity of the network, the probes and DPI boxes also need to be upgraded. This means operators need to spend more whenever they expand their network.
Correlating Packet Broker
Packet Broker connects taps/span ports to Probes. A Packet Broker provides additional value to operators in reducing incremental "Visibility-Tax". A Smart Packet Broker understands the packets they are brokering. Smart Packet Brokers do not just load balance the traffic to Probes, they can "Correlate" both Control (GTP-C) traffic and User Plane (GTP-U and SGi) traffic. By decoding the GTP-C traffic and tracking subscriber session, a Smart Packet Broker can stick the GTP-C, GTP-U and SGi traffic into the same probe throughout the session irrespective of Inter Radio Access Technology (IRAT), Mobility, Handover scenarios and internal link-level load balancing that might otherwise spray the packets across several probes.
In the absence of Correlation at Packet Broker level, subscribers' traffic would end up processed by several probes and that calls for second level of aggregation or correlation, which are typically implemented at the analytics platform. Correlation feature not only reduces the compute needs of Analytical layer, it also optimizes the Probe .
As Gi throughput grows, operators need to invest in additional DPI boxes for usage analytics and for security purposes. A Smart Packet Broker can greatly save cost by dropping known, good traffic. For example, an operator who is deploying DPIs for scanning of HTTP traffic for malware and bots can avoid sending streaming video traffic. Similarly if GTP-U traffic were to be dropped based on domain/application, it needs to be done based on inner IP. A Smart Packet Broker brings the capability to manipulate packets based on inner IP in order to reduce volume of packets hitting Probes.
In the coming blogs I will talk about more critical features that are useful for mobile operators.