Byte-level caching focuses on the entire data traffic path: from the device through the RAN and backhaul, and into the network core.
For mobile operators, it’s the best of times and the worst of times. According to Yankee Group, mobile data traffic is growing more than 60 percent annually, but average revenue per user (ARPU) growth is in the low single digits.
One option is to increase tariffs, but that will increase churn. Another option is to throttle the heaviest users and services, but that could create a backlash from customers and regulators.
A third option enables operators to manage bandwidth and optimize data without negatively affecting the customer experience.
Known as “byte-level caching,” this approach focuses on the entire data traffic path: from the device through the RAN and backhaul, and into the network core.
Before data is transmitted over the network, byte-level caching steps in to replace repetitive bytes with shorter tokens. The technology works with any airlink and backhaul technology, such as TDM and LTE, and any protocol, such as TCP. It also works with nearly any type of data traffic where repetitive bytes can be cached rather than continuously retransmitted across the network, including VoIP. This broad approach gives operators more optimization options compared to alternatives, such as Web and file caching, which address only one part of the traffic flow.
Operators use byte-level caching for 3G and 4G, because it has a greater effect on the bottom line and the customer experience than if it were applied to 2G and 2.5G. Yankee Group’s recent analysis of byte-level caching included looking at the amount of traffic reduction operators could reasonably expect. The firm found that byte-level caching could be applied to about 60 percent of 3G traffic (for a 30 percent reduction in total traffic) and 40 percent of 4G traffic (with a total reduction of 30 percent).
Those reductions flow straight to the operator’s bottom line. For example, a typical Tier 1 operator with a mix of 4G and 3G customers can save up to $800 million over five years in backhaul costs alone, according to an ROI calculator that CENTRI jointly developed with Yankee Group1. Additional savings are possible if that operator doesn’t have to purchase more spectrum, add thousands of cell sites, or both. As a result, the operator now has a lower cost structure, so it can price its services more competitively, yet profitably.
Byte-level caching also helps minimize churn, which Yankee Group says cost U.S. operators a total of $25.7 billion in 2012. The firm surveyed more than 2,000 U.S. mobile users to identify which service attributes influenced their decision to stay or churn. Not surprisingly, low prices ranked highest, which byte-level caching addresses by lowering the operator’s CapEx and OpEx.
Byte-level caching also enables operators to provide a consistent network, which respondents said was the No. 2 factor affecting their decision whether to churn, as well as operators’ ability to provide the fastest network, which was the fourth factor.
Staying Ahead of the Curve
For operators, the alternative to byte-level caching is to get more money from investors and customers to buy more infrastructure, backhaul, and spectrum – an option that has obvious drawbacks. For one, spectrum is in perpetually short supply, a situation that’s unlikely to change anytime soon. The FCC’s National Broadband Plan recommends making an additional 300 MHz available for use within the next five years and 500 MHz within 10, while Infonetics Research2 says the need is closer to 1,000 MHz within five years. U.S. mobile operators currently have access to about 300 MHz, so achieving those targets means freeing up two to three times more.
Even if money were no object, operators have the challenge of mobile data traffic’s spiky nature, which is as much a problem as traffic growth. Yankee Group’s interviews with Tier 1 mobile operators found that the peak-to-normal-traffic ratio in the core is 12 to 1.
“Mobile network operators (MNOs) realize that they must leverage a variety of tools to cope with data traffic growth,” concludes Yankee Group. “The ability to smooth out traffic peaks, as well as reduce overall traffic volume by as little as 10 to 20 percent, can enable an operator to defer an infrastructure upgrade (i.e., from 40 Gbps routers at the core to 100 Gbps routers) for 12 to 18 months. For a large carrier, that can equate to hundreds of millions of dollars in savings.”
Many operators claim it won’t be long before demand outstrips their ability to make upgrades. AT&T is one example. “The capacity we carried in 2008 will be a rounding error five years out,” says John Donovan, AT&T's senior executive vice president of technology and network operations. “Our 2 Gbps backbone lasted seven years. Our 10-gig lasted five. Our 40-gig will last three. You get to 100 gig, what’s that, 18 months? At 400 gig, I think routers melt. The finance [department] likes liquid assets, but I don’t think that’s what they have in mind.”
Mobile operators can’t predict the next Facebook, iPhone, Netflix, or Snapchat. That means they can’t add enough infrastructure, backhaul, and spectrum in advance so they’re ready when a new device or service appears and sends traffic soaring. In the past, operators had tight control over the devices and services that used their networks. No more. Today, device vendors, application developers, and over-the-top (OTT) service providers have the most influence over the types and amounts of mobile data traffic.
That new paradigm is here to stay, which is yet another reason why mobile operators see byte-level caching as the most effective way – in terms of cost and flexibility – to prepare for whatever hits their networks. As Yankee Group recently put it: “To succeed, MNOs must prepare for what they cannot predict, but must do so in a cost-effective manner. By embracing solutions that optimize network performance, MNOs play directly to the criteria most important for customer retention. They control their own costs, improve network consistency and, by doing so, enhance their brand. Byte-level caching is one of these essential solutions.”
This article originally appeared in the January/February print issue. Click here to read the full issue.