Is it live or is it Memorex?

By Jeff Finkelstein

The cable industry has been hearing about the magical mystical virtual CMTS for the past 14 years or so.

The cable industry has been hearing about the magical mystical virtual CMTS for the past 14 years or so. The transformation of the CMTS from a monolithic to virtualized architecture has been in progress for the past ~10 years. At the same time, companies have been transforming their internal networks from single-purpose highly-optimized devices to a virtualized environment using general purpose servers. The over-used term to describe the virtualized world the IT groups want to use in many cases is “Common Platform,” meaning a single platform which will be used for all virtual services including OSS, BSS, NMS, EMS, web servers, database engines, and even networking. It is irrefutable to me that making the transition to virtual CMTS is not only logical, but essential to the long-term viability of the network, especially with the reality of legacy CMTS architectures no longer being made and soon not supported.

As someone who started in real-time computing and networking, I have many questions about the use of general purpose computing for mixed services, especially those mixed with network services. The most important one being “does a Common Platform work for back-office AND networking?”. While there are many common functions, the ones that concern me are the inherent differences between real-time and non-real-time computing. Not that things cannot be made to work in a Common Platform, more that because it can does not necessarily mean it should.

First, let us start with a comparison of functions…

Comparing real-time and non-real-time computing

Computing systems are integral to modern life, powering applications from scientific research to financial markets and entertainment. However, not all computing systems operate under the same constraints or serve the same purposes. A fundamental distinction exists between real-time and non-real-time computing. We need to explore their differences, advantages, disadvantages, and real-world applications, to fully understand the challenges and why vCMTS is a far more complex beastie than many believe it to be.

What is real-time computing?

Real-time computing refers to systems where the correctness of computations depends not only on their logical accuracy but also on their timing. In other words, a system must produce outputs within strict time constraints to be considered real-time. Such systems are often used in scenarios where delays could result in catastrophic consequences or degraded performance.

Characteristics of real-time systems

What makes a real-time system so different?

1. Determinism: The system’s behavior must be predictable. Inputs processed at the same time under identical conditions should yield the same result within the specified period.
2. Low Latency: Responses must occur within a few milliseconds or even microseconds in some applications.
3. Reliability: The system must operate consistently and meet timing requirements under all defined conditions.
4. Prioritization: Real-time systems often use task prioritization to ensure critical operations are completed first.

Types of real-time systems

1. Hard Real-Time: Missing a deadline is considered a system failure. Examples include medical life-support devices, anti-lock braking systems, aircraft control systems, and network responsiveness.
2. Soft Real-Time: Deadlines are important but not absolutely critical. Performance degrades if deadlines are missed but does not lead to catastrophic outcomes. Examples include video streaming and online gaming.

What is non-real-time computing?

Non-real-time computing systems, by contrast, are not bound by strict timing constraints. Their primary goal is to maximize throughput, efficiency, and accuracy without being tied to deadlines. These systems are typically used in scenarios where response times, while important, do not need to occur within milliseconds or microseconds.

Characteristics of non-real-time systems

The main characteristics of non-real-time systems are:

1. Throughput-Oriented: These systems aim to handle large amounts of data or computations over time. For example, using a database engine to construct datasets for processing.
2. Flexibility: They can tolerate variations in execution time without significant consequences.
3. Scalability: Non-real-time systems often scale more effectively across distributed architectures.
4. Fault Tolerance: While reliability is important, these systems can often retry operations or queue tasks without catastrophic failure.

Examples of non-real-time systems

• Batch processing systems
• Web applications (e.g., e-commerce websites)
• Big data analytics platforms
• Office productivity software (e.g., word processors)

Advantages of real-time computing

For specific applications, real-time computing has a significant advantage over its less strict relative in non-real-time computing.

1. Predictability: Real-time systems provide consistent and reliable outputs within defined timeframes, essential for mission-critical applications.
2. Safety: Systems like pacemakers or airbag controllers ensure human safety by meeting stringent deadlines.
3. Efficiency: Optimized scheduling mechanisms ensure critical tasks are prioritized.
4. Specialized Hardware: Many real-time systems are paired with hardware optimized for low latency and high reliability.

Disadvantages of real-time computing

As well, there are some disadvantages that must be considered.

1. Development Costs: Designing and testing real-time systems is resource-intensive.
2. Limited Flexibility: These systems prioritize timing over adaptability to changing requirements.
3. Complexity: Real-time systems require precise synchronization and specialized development environments.
4. Resource Usage: High-priority tasks may monopolize system resources, leading to inefficiency in handling less-critical tasks.

Advantages of non-real-time computing

Non-real-time computing is useful for many applications, just not those that require real-time response.

1. Cost-Effectiveness: Non-real-time systems are generally less expensive to develop and maintain.
2. Flexibility: These systems can handle a wide range of applications and changing workloads.
3. Scalability: They adapt well to distributed and cloud-based architectures.
4. Efficiency: Focus on throughput and resource optimization makes them suitable for handling large-scale computations.

Disadvantages of non-real-time computing

1. Unpredictable Response Times: Variability in task completion time can lead to suboptimal performance in time-sensitive applications.
2. Lower Priority Handling: Critical tasks may be delayed in favor of less important ones.
3. Latency: The lack of strict timing constraints can lead to delays that are unacceptable in certain scenarios.

Real-world applications

It is easy to think where you may use real-time vs non-real-time systems, but it is not quite as black and white as some believe.

Real-time computing

1. Industrial Automation: Real-time systems control robotic arms, conveyors, and other machinery, ensuring precise operations.
2. Medical Devices: Systems like ventilators, infusion pumps, and heart monitors rely on real-time computing to save lives.
3. Transportation: Air traffic control, autonomous vehicles, and train signaling systems depend on real-time responses.
4. Telecommunications: Real-time systems manage network traffic to ensure minimal and consistent latency with uninterrupted communication.

Non-real-time computing

1. Data Processing: Big data platforms like Hadoop or Spark process massive datasets without immediate time constraints.
2. Content Delivery: Platforms like Netflix and YouTube stream video content, buffering data to manage latency.
3. Scientific Research: High-performance computing clusters perform simulations and analyze data without the need for real-time responsiveness.
4. Business Applications: Enterprise resource planning (ERP) systems, customer relationship management (CRM) software, and other back-office tools fall into this category.

Challenges in blurring boundaries

The evolution of technology has blurred the line between real-time and non-real-time computing. Many modern systems incorporate elements of both paradigms. For instance, an online video game might require real-time responsiveness for gameplay while performing non-real-time background analytics to recommend items or optimize server performance.

This hybrid approach introduces challenges:

1. Complexity: Designing systems that integrate real-time and non-real-time components increases architectural complexity.
2. Resource Allocation: Balancing resources between time-critical and non-critical tasks can be difficult.
3. Reliability: Ensuring that real-time components function correctly without being affected by delays in non-real-time operations requires careful planning.

The future of computing

Advances in hardware and software continue to expand the capabilities of both real-time and non-real-time systems. Emerging trends include:

1. Edge Computing: By processing data closer to the source, edge computing bridges the gap between real-time and non-real-time requirements, reducing latency.
2. Artificial Intelligence: AI-powered systems are being deployed in real-time applications like autonomous vehicles while also driving non-real-time analytics in industries like marketing.
3. 5G Networks: The low-latency capabilities of 5G enable new real-time applications in areas such as telemedicine and augmented reality.
4. Quantum Computing: While still in its infancy, quantum computing holds promise for revolutionizing both real-time and non-real-time computing by solving complex problems faster.

Conclusion

Real-time and non-real-time computing serve distinct purposes, each with its own strengths and limitations. Real-time systems are indispensable in scenarios requiring strict timing constraints and high reliability, while non-real-time systems excel in flexibility, scalability, and cost-effectiveness. Understanding their differences is crucial for selecting the appropriate approach for a given application, and ongoing technological advancements continue to enhance both paradigms, enabling new possibilities and hybrid solutions.

Simply put, while we would all like to use a single virtualization platform that meets all needs, just because you could does not mean you should. The requirements of real-time computing, especially with virtual CMTS and routers, is not the same as web servers or other non-real-time processes.

 

Images provided by author, Shutterstock

---

Jeff Finkelstein,

jlfinkels@gmail.com

Prior to retirement, Jeff Finkelstein was the Chief Access Scientist for Cox Communications in Atlanta, Georgia. He has been a key contributor to engineering at Cox since 2002 and is an innovator of advanced technologies including proactive network maintenance, active queue management, flexible MAC architecture, DOCSIS 3.1, and DOCSIS 4.0. His responsibilities included defining the future cable network vision and teaching innovation at Cox. Jeff has over 50 patents issued or pending. He is also a long-time member of the SCTE Chattahoochee Chapter and member of the Cable TV Pioneers class of 2022.