Abstract
Cloud services in the telecom industry is facing transitions. Currently, two paths have been placed ahead of operators with the gradual growth of technology: public cloud or telco cloud for core network deployment. This article provides an overview of the public cloud adoption for operators. Compared with telco cloud service, the extensive use of public cloud may face many challenges in the following aspects:
The data security of public cloud service concerns the public about its underlying logic and trust issues, and there is currently no mature division of data sovereignty.
Considering total lifetime, the public cloud may generate more costs considering its complex charging methods.
As for the problem of public cloud service to meet the customization requirements of the core network deployment, especially in real-world practices, It is still difficult to solve it in the short term.
The inadequateness of meeting an operation at the telecom level should be considered. It may lack telecom level availability and a responsible entity. The Available Zone (AZ) deployment structure takes more advantage of complexity to operators at this time.
Introduction
Cloud services in the telecom industry is facing transitions. Previously, telecom operators usually deploy their core network at privately hosted servers. With the gradual growth of the 5G technology, clouds service deployment is much more prevalent these days. The technology of 5G empowered the structure of cloud-native to be more consistent and more approachable. Step by step, two paths have been placed ahead of the telco cloud service and user companies——whether they should choose public cloud or telco cloud to be a major way of deployment.
There are emerging players in the public cloud fields. Amazon Web Services and Microsoft Azure, two leading multinational corporations in the cloud industry, both have announced their innovative decision of working with large telecom operators in public cloud deployments. The public cloud’s feasibility in core network deployment gradually becomes a key focus in the telecom industry. Empowered by 5G, this service is generally considered to be fully decoupled, flexibly used, and low costing. However, some of its inherent disadvantages, such as the data security concerns, future costs, and standardized design, are always being challenged by the public. On the other hand, telco cloud service, which is a more customized way of cloud construction, is growing steadily and has more practical applications. Still, the future of cloud-based structure in the telecom industry is undecided.
We may not expect a sudden leap into the transformation procedure of the public cloud service. The future of cloud-based structure in the telecom industry is still undecided.
- Data security
The data security of public cloud service concerns the public about its underlying logic and trust issues, and there is currently no mature division of data sovereignty.
Public cloud data security concerns operators in terms of its structural design and data sovereignty. Data security issues in cloud service attract great attention around the globe. Indeed, the public cloud service’s security is often questioned. There appears to be a great risk of vital data leakage, as the data is being stored and managed together in public infrastructure. Compared to telco cloud, there appears to be way too much uncertainty in the whole data transmission and storage procedure.
In nature, the public cloud provides an escrow service built on technological advances and trust. Let’s first look at the underlying logic of this escrow service. The public cloud company safeguards data and documents for different companies altogether. This naturally leads to the concern of data clutters, hijacks, and so on. Data leakage may happen at any time in the whole data transmission procedure, while a large-scale public server can easily become a hijack target. What’s more, when the operator needs to transfer its data from a previous public cloud server to a new provider’s server, the subsequent treating procedure of the data leftover on the previous server remains a black box to operators. As regard to trust issues, security concerns are not groundless. Internal staff may have the access to edit and transfer operators’ data. Whether the internal security control is efficient or not, the debate over public cloud security has never stopped. In the recent press release, Alibaba Cloud has been accused of disclosing user information by internal staff to a third party without the consent of users. This, to a greater extent, aggravates users’ concerns about the security issue.
Meanwhile, data sovereignty under the public cloud service structure is ambiguous. Data sovereignty, a focus of policymaking and legislation in many countries, would post higher requirements on multinational companies especially. Russia has been working on the legislation and execution of data sovereignty since a few years ago. The Chinese government is also working on the data sovereignty issue recently. The current execution status, which urges companies that collects citizens’ information to store those data inside the country, requires a quick response from cloud companies. The deployment of public cloud is standard and consistent, which means all the infrastructure part is built and regulated together by a certain company in a certain country. In many cases, it doesn’t satisfy the required local deployment. Local deployment is indeed a solution to domestic data control and regulation and some leading multinational cloud service providers, like Amazon Web Services, could conduct local deployment across countries. However, examples like these are rare. In the real case, local deployment isn’t a common agreement. Operators who have uploaded all information to the public cloud may be more vulnerable to threats to a certain degree.
- Cost and maintenance
Considering total lifetime, the public cloud may generate more costs considering its complex charging methods, while a telecom level SLA may not be guaranteed under low-cost design.
TCO of public cloud is ambiguous, and it may generate more and more costs as time goes by. From a popular point of view, public cloud service seems cheaper at first glance. Almost free from OPEX in the beginning stage, free-to-go cloud service is attractive to small and medium-size company users, for its accessibility and affordability. This view, however, does not consider the size and lifetime of the business. In the CAPEX part, the public cloud service has a complex set of charging methods. The construction of public cloud services requires adequate computing resources and bandwidth, which may generate high capital-output. In a privately designed telco cloud, users have full access to their data, while most public cloud companies follow the rule of ‘more fees, more function’. It means users would need to pay more for more access permission, more storage and more function. This would generate a huge hidden cost, and it would grow gradually as time goes by. Telecom operators have typical requirements for long-term cloud use. For them, consistency over time is of great importance. The overall performance of public cloud cost remains undecided concerning possible future expenditures.
Theoretically, the public cloud relies on the ‘over distributed mode to ensure its low costs. Over-distributed means resource sharing to more end-users is made available through unbinding CPUs and peak runs. This method enables lower OPEX through cost apportionments. However, this design could be fragile during the high-capacity period, not to mention the disaster capacity required by the telecom level Service Level Agreement (SLA). A telecom level SLA is not guaranteed, and the computing power is unstable. For telecom operations, CPU resources should be sufficient, and it should not need a race to control. So, if public cloud servers decide to present a cloud service that is practicable with the telecom level, the extent to which cost reduction can be reduced is questionable. Considering these two aspects, it is hard to say which costs are cheaper.
- Customization
As for the problem of public cloud service to meet the customization requirements of the core network deployment, especially in real-world practices. It is still difficult to solve it in the short term. The learning barrier of switching to the standardized public cloud service is also high.
Public cloud service is incapable of meeting the customized requirements of operators. Most companies have exclusive regulations in data management, storage and differentiated service. Typically, internet companies, or other companies whose businesses are strongly related to data, usually have very customized needs in data services. Different countries and districts also have different rules in terms of data management. As to the service side, operators are offering customized service to customers’ end. In this case, the public cloud may not be capable of meeting all these customized needs.
Essentially, all functions and solutions provided by public cloud service companies are standardized, which means that they are not specifically designed to fit in some customized needs for one single user. Indeed, public cloud service companies are trying to offer different layers of function parts that attempt to fit various types of consumer’s needs, though this would result in high costs in human resources, materials and time in this manual adjusting procedure. Operators do not need to rebuild a fresh cloud structure. Instead, they may freely choose from different function parts that cloud service companies could offer and combine them into a less standardized product that would match the company’s needs. A representative example is Amazon Web Services has developed a block-based solution, which means they segment their public cloud products into different small function blocks. Users could freely choose from different blocks and assemble them into a specific product that aligns more closely with the company’s needs. Other companies are following this trend, but their performance doesn’t have many existing references.
However, even a more customized construction can be theoretically achieved in the public cloud service field, in the real telecom industry practice, there is more suddenness that requires real-time customization and rapid response, which may be somewhat beyond the scope of preparation. For example, in emergency network conditions, the more customized the system is, the quicker the response would be. This apparently, posed a huge challenge for public cloud service participants.
Moreover, although most public cloud companies are trying to optimize their product by lowering learning barriers, the learning barrier is still considerably high, compared to traditional private design. For companies who are used to conducting data storage in a more customized system, it takes so much time to get familiar with a new and standardized cloud product. The IT department of the user’s company, which may lack proficiency in new cloud technology, shoulders great responsibility in the public cloud adoption, which may drag the user company into risky conditions. Overall, the public cloud still has a long way to go in building a more customized system.
- Availability
The inadequateness of meeting an operation at the telecom level should be considered. It may lack telecom level availability and a responsible entity. The Available Zone (AZ) deployment structure takes more advantage of complexity to operators at this time.
For the telecom industry, an operation at the telecom level, which means high availability, high accuracy, high capacity, and low latency, is crucial. Compared to telco cloud, it is harder to maintain a public cloud system at a telecom level. The fully decoupled environment of the public cloud service brings about a much more complicated internet structure. We may figure out more differences between public cloud and telco cloud by examining their carrier-class SLA.
Firstly, service availability is an important indicator for telco services. It refers to the percentage of time that a system can operate normally. Availability for service not only gives guarantees that data is not lost but also ensures continuous availability for service. Usually, service availability in the public cloud industry is generally 99.9% or 99.99%. In terms of telco cloud, it should be operated at 99.999% availability (Five-nines availability) due to its unique application scenarios, which promises users a service interrupted session of only 0.864 seconds per day, that is approximately 5 minutes per year.
Secondly, public cloud service suffers more for lacking a responsible entity. This is an inherent flaw in the structural design of public cloud services. Public cloud companies are only responsible for the operation and maintenance in the Infrastructure as a Service (IaaS) layer. Splitting functions and responsibilities would result in low consistency in connections. Therefore, the public cloud could not achieve true operation and maintenance automation. When it comes to unexpected situations, such as a network failure, users are easily confronted with difficulties in maintenance. Such conditions are less likely to happen in a network privatization service, where a single party takes charge of all effectively. More professional people are needed in the whole public cloud’s maintenance service, which may be difficult to achieve in the current situation. It requires more collaboration and education about product use in the long run.
Thirdly, the deployment structure of the public cloud relies on the region and Available Zone (AZ). AZ is an independent physical machine room in one region. There are usually various AZs in one region, which are used to ensure the normal operation of other AZs if one AZ fails. This circumstance would bring more complexity to internet construction, and more availability risks would occur. In a more common situation when there is no locally deployed AZ, operators would need to build a new AZ through some special type of connection. Things will get far more complicated. Moreover, the operator might lose control of the network topology. Operators may only rely on cloud service companies in network building and dilatation.
Conclusion
In conclusion, for the telecom industry, we believe that telco cloud has comparative advantages in the trend of cloud transformation at present. The advantages of telco cloud lie in its high-security, high-reliability promise, and customization capability. These two abilities directly answer the operator company’s biggest concerns. Compared to it, the extensive use of the public cloud still faces many challenges. It is still unclear whether these challenges could be solved. We may not expect a sudden leap into the transformation procedure of the public cloud service.
