Are Telecoms Vendors Addressing the Commercial Opportunity from Consumer-Specific SLAs in Distributed Networks?

Mobile Service Providers’ (MSPs) positioning in the global production frontier remains anchored to hard-to-duplicate network assets and infrastructure that continues to yield profits. Existing revenue streams notwithstanding, MSPs are incessantly seeking to pursue further growth, particularly with 5G, which introduces innovation on the technical and commercial fronts. Investment in the core network and cloud native architectures are expected to unlock many 5G-related commercial opportunities on the consumer and enterprise fronts. One such opportunity is endpoint Service Level Agreements (SLAs), the ability for MSPs to accurately assess what a failure’s impact is on consumers, an engineering feat that is commonplace in the Business-to-Business (B2B) space.

SLA provisioning in the B2B domain is attributable to two reasons: one, the B2B segment has always been much more SLA-driven, with strict metrics in term of planned and unplanned maintenance, and two, a specific network and/or hosting resource is easily associated with a customer, as opposed to a consumer who is characterized by high mobility. The opportunity at stake will become more pronounced as MSPs deploy 5G core networks that natively support low latency and high bandwidth capabilities. A discussion about what node-specific SLAs entail can be framed around two lenses: one, the realities of current compute ecosystem in telecoms; and two, network specifics particularly when viewed from a decentralized, distributed model such as the Internet.

MSPs are in full swing in terms of shifting their Network Functions (NF) to software appliances. In practice, there are three options to implement a NF, all of which need to be considered from an assurance and SLA perspective. A NF can be implemented as an element on a dedicated hardware, as a software instance running on hardware (virtual network function, VNF), or as a virtualized/Cloud-Native function (CNF) running on cloud infrastructure. This heterogeneous compute environment presents a service-resource relationship that is no longer based on a foundation with a 1:1 mapping. For orchestration and management solution providers, such a fluid arrangement of NFs and compute resources necessitates control over two dimensions: a) management of NFs, and b) management of the virtualization layer of NFs. Inevitably, there will be a marked effect on service assurance and Quality of Service (QoS) requirements that vendors ought to consider if they are to assess the impact of a failure at the node level.

When we think of a perfectly assured network, we are presented with numerous discussion points. The networks of today are built from a variety of transmission media, including wire, cable, fiber, and wireless radio technology. The resulting functionality and performance that will need to be considered for node-level SLAs is affected by all those components. Therefore, as shown in Table 1, ecosystem heterogeneity and openness are key determinants of exactly pinpointing a potential endpoint failure. Whether it is a hardware appliance or a virtual one, distributed compute or centralized resource pool, MSPs are compelled to put together a set of dispersed and distinct network functions and elements to establish a service. The ability to assure that an End-to-End (E2E) service is of paramount importance in order for the industry to unlock the opportunity at stake.

E2E service assurance and performance management in hybrid networks constructed from heterogeneous hardware and software may be achievable. However, ecosystem openness, which allows components to be added or replaced, must be in place for prospective commercial offerings to enable service uptime guarantees at a consumer level. Today, there is no tightly coupled integration between telco support systems and service assurance. There are processes in network operations that are fully integrated and operate in real time, but there are others that remain static and isolated. It is the latter that can potentially create back doors that can give rise to (invisible) dependencies between apps and workflows. An iterative and automated approach is therefore required to bring together the performance and analytics capabilities that permeate today’s isolated, top-down network deployments predicated on manual processes and scripting.

In the past, network engineers could walk into a MSP’s data center and physically troubleshoot and remedy failures. Today, the net result of virtualization is that software instances are no longer bound by locality, since they could be running in cloud/data center environments that are geographically dispersed. The challenge for vendors is to integrate those different locations together for an E2E workflow of the consumer service. Furthermore, today’s operations maintenance is conducted in an open loop fashion. Existing solutions provide some degree of visibility, which is used to dispatch a network operator to the faulty site. Network healing, moreover, is predominantly manual, with the sequence of remedial steps considered as tacit knowledge not able to be codified. To that end, almost all vendors are incorporating intelligence and analytics into their assurance products so that incremental steps can be taken toward closing the loop in an automated fashion. Accedian and TEOCO are two vendors, among many others, that utilize a purpose-built engine to ingest only “clean” data at ingress traffic and reject superfluous data, thereby cutting down the noise.

A profound question that a fluid resource environment raises, and one that vendors will need to grapple with, is network interface identification and location addressing. Networking as we know it today does not separate the notion of identity from the notion of location; in other words, the identity used by a device is also the location where traffic will ultimately end up. In software-defined architectures, network services will no longer be tightly coupled to the underlying infrastructure. Routing traffic to a fixed location in a network where nothing is really fixed raises new requirements that ought to be addressed if the commercial potential from node-specific SLAs is to take off. In addition to the Internet Protocol (IP) architecture based on the location (i.e., the “where”), a virtual world requires that vendors factor in knowledge of resource identity (i.e., the “what” or the “who”).

In addition, today’s telco networks mirror a synchronous distributed system. In other words, high reliability (99.999%) can be provided because each message is transmitted over a connectivity medium within a known bounded time and each process performs tasks with known resource requirements. The Internet, by contrast, is an asynchronous distributed system that allows no assumptions for the time intervals involved in any process execution and message transmission. There may be times when a message is delivered from process A to B within an agreed SLA, but there may be instances when a message reaches its destination after an arbitrarily long time. This, in a nutshell, is the challenge that vendors such as Ericsson, Huawei, Nokia, and ZTE must conquer from a technical perspective.

At present, there is no vendor that enables MSPs to define an SLA down to the person that is consuming an application in a consistent and elegant way. What does the industry do when the tech ecosystem is so complex and the ability to monitor it is so limited? Absent extreme product openness, one option may be for MSPs to restrict the number of solution providers in order to reduce the range and complexity of the network they have to monitor and manage. But, if the industry at large collectively works to institute an ecosystem with uniform business connections among different entities (infrastructure vendors, software vendors, and open source community), create a data and AI telco standard, and draw significant management buy-in, node-specific SLAs may well turn out to be the next wave of growth for both the demand and the supply sides in telecoms.