My name is not Morpheus and I’m not asking you to choose between a blue or red pill here, but let me start off by saying I started out as a hardware guy in the business. I know hardware, I like hardware but I have converted into being a software dude. Yes, the Holodeck is amazing as was the Matrix but they didn’t do it for me. The power of software is amazing and we all see and use it every day, even those of us who are not in the cable business. Software is the way forward with some very powerful needed help from smaller hardware devices. Of course, we will always need hardware but not in the same way and magnitude we use it today in our industry. Software can now, in many ways, do or mimic what lots of hardware used to do. Getting the bugs out literally used to mean getting “bugs — real moving ones” out of mainframes as large as most homes.  Virtual cable modem termination system (CMTS) or converged cable access platform (CCAP) equipment will transform the way our industry manages and deploys DOCSIS over the coming years — and it’s a good thing. We need to compete with fiber and telecom operators and a way to get the cost down is by using software (virtualization, PNM and CCAP) and less hardware. The Holodeck  might be in the future still but the industry is ready for virtual CMTS-CCAP.Now there will be some skeptics. However, a great idea is hard to stop; just look at what Star Trek did for us. Wikipedia defines virtualization as the act of creating a virtual (rather than actual) version of something, including virtual computer hardware platforms, storage devices, and computer network resources.  Often it can be confusing to understand just what us technocrats are talking about (fancy way of saying nerd) when discussing virtualizing servers and even more so when we start talking about virtualizing CMTS-CCAP. What does this mean and why would anyone do it? Here is why. I will cover the concept of virtual CMTS-CCAP, how it pertains to DOCSIS and why this is a disruptive technology in our industry. I recommend everyone understand the basics if you are not already looking at it.
I’ve first wrote about software and virtualization  back in 2014. I will cover the basic concept of virtualization for computers (i.e., servers). Every computer must have an operating system (OS) running on the hardware. The operating system can be Windows, Linux, Apple OS X, etc. This is necessary to interact with the computer’s hardware and accomplish work, such as writing the article you are reading. Virtualization can trace its roots back to the 1960s, where it was used to logically divide tasks across large mainframe computers due to the significant costs of the mainframes. 
Today, virtualization is pervasive on single user computers. A common use is to simply run two operating systems simultaneously on the same computer. It’s likely you know someone or you are using virtualization to run a combination of Windows, Linux or OS X on the same computer at the same time. That is virtualization and running an operating system in this manner is called a virtual machine or VM for short.
Datacenters will commonly deploy more expensive servers, equipped with a lot of RAM, hard drives and processors, then use VMs to run more than one operating system on a single server. Figure 1 shows the migration from a single server running a single OS to server virtualization running many OSs.
There are numerous benefits to doing this, of which I’ll list just a few. First, you have one server running many operating systems (VMs). This reduces space, power, heat, etc. Second, virtualization has a cool property called snapshots. Once your VM is configured and running properly you take a snapshot of the VM. If something bad occurs with the VM, for example, you get malware and your entire operating system becomes corrupted, you can restore your VM back to its pristine condition in a matter of minutes or even seconds using the snapshot you made — like the disaster never even happened. Next, snapshots of your VM can be moved from one server to another. This means if your server has problems or dies completely, you can be up and running on another server in just a few minutes using the snapshot of your VM.
The power of virtualization should now be a little clearer. You can run multiple operating systems on the same computer. You can take snapshots of the operating systems at any time and restore these snapshots at any time in the event of a disaster. You can backup these snapshots and restore them to a completely different server at any time, which is extremely important when hardware fails. But what possibly could we do with virtualization and DOCSIS?
Applying virtualization to DOCSIS is extremely transformative and will change the way our industry is able to manage and deploy DOCSIS networks over the coming years.
Since the beginning of DOCSIS we have deployed CMTS-CCAPs in a non-virtualized way. The operating system of the CMTS-CCAP runs directly on a large chassis that is analogous to a mainframe system. The CMTS-CCAP itself is large, generates immense heat, draws substantial power and has many points of failure. While these “mainframe systems” have many redundancies, they are still large, power hungry and inflexible.
When you purchase a CMTS-CCAP it could potentially become obsolete with a new DOCSIS specification or you will run out of capacity due to subscriber demand. Hardware-based CMTS-CCAPs are simply limited in many ways. Most people are familiar with the term fork-lift upgrade, which means the old CMTS-CCAP gets physically replaced by the new model.
When new software is available for the CMTS-CCAP, upgrading may require the entire CMTS-CCAP be taken offline including all subscribers. If there is a bug in the software and it must be downgraded to the older version, there is no snapshot process as with virtualization. We must again take all users offline and install the older software. For this reason, most operators have a lab system to test out new software updates for bugs.
The hardware and software are closely integrated. Making a change to the CMTS-CCAP hardware, such as upgrading from DOCSIS 3.0 to DOCSIS 3.1 requires substantial changes to the software. Often even small changes in the CMTS-CCAP operating system can have unintended consequences to other areas, such as the monitoring systems which communicate with the hardware via SNMP.
Cable operators and hardware vendors have recognized the potential of virtualization and have produced early stage products that are virtualized CMTS-CCAP systems. The concept is simple; the CMTS-CCAP processing component that performs the heavy lifting of routing traffic and managing modems is moved into a virtualized environment — this is just a server running an operating system on a virtual machine. Nearly all major vendors have various solutions and new companies are also entering the market.
There are some obvious benefits of having your CMTS-CCAP running on a 1U server, such as all the headend space that is recovered. Others include the substantial savings on power, cooling and strained backs from installing heavy equipment. These are just the tip of the proverbial iceberg. Remember why we use virtualization for computing — in theory each service group (or fiber node) could have its own VM on the virtual CMTS-CCAP. Each VM could be running its own software version. Why would we want to do this? For one, if there were an issue with one service group — be it an upgrade, capacity, security, an outage, you name it — we are now able to treat each service group like it is running on its own CMTS-CCAP. We can upgrade it, downgrade it, reboot it, apply special security rules to it, different SNMP community strings — I can go on and on, but I’ll let you use your imagination because we are only limited by the developers. This becomes really exciting with how far we can go.
Also, think about snapshots. These are also possible. Gone are the days where we are afraid to touch the CMTS/CCAP. If we make a mistake we just roll back the snapshot to the last working state.
Scalability is interesting to think about. When you need to add more service groups or capacity you simply add another server — licensing will of course apply, but that is assuming your vendor wants to make money. Similarly, redundancy is accomplished by adding another server. Many things become simplified in the same way that we think about networking and maintaining computers in a datacenter. What’s more is that we can leverage our existing IT resources to support this infrastructure.
The question on many people’s minds is “servers don’t have RF connectors.” So how do we get RF in and out of the server? This is the interesting part. Modular CMTS (M-CMTS) and remote PHY (R-PHY) architectures have created the building blocks. Edge-QAM modulators and R-PHY devices have an Ethernet port for an input and RF ports as an output. This allows us to connect the virtual CMTS-CCAP to our beloved RF connectors and get QAM and other RF signal output/input.
Because this solution is inherently so flexible I foresee vendors being able to offer remarkably diverse solutions based on virtual CMTS-CCAP platforms. The core would always start with their virtual [DOCSIS] software, which could reside on a server in a hubsite, the headend, a datacenter owned by the operator or even a datacenter owned by someone like Amazon or Google. Next, vendors can offer al-la-carte solutions to meet the specific needs of the operator. In some areas, it may make sense to put all the RF in the fiber node while in others only the return RF will be in the fiber node, while keeping the downstream RF in the hubsite. In urban areas, we could see solutions where all the RF is deployed directly in MDUs.
Now let’s pull all of this together with a reminder that CCAP was designed to be flexible in supporting the growth in QAM channels and its capabilities have only been expanded with DOCSIS 3.1. With virtual CCAP vendors could offer cable operators an amazingly flexible system that is not only cost effective and resilient but also has the amazingly cool virtualization features I’ve already mentioned, but they can go one step further. With virtual CCAP the system licensing could be based on dynamic capacity usage. The concept is simple — the operator only pays for licenses based on the QAM or OFDM signals required at any given time — this is a pay for what you use model.
So, there are a lot of options for cable operators. Free up real estate, save money on power consumption, pay for what you use and easily add what you need. Again, you will always need some form of hardware. That form of hardware will change, though. Virtualization makes things possible that only seemed like science fiction years before. It’s an exciting time to be in the industry to see all the changes going on and endless possibilities.
Software, although less expensive than buying a lot of hardware, still has expenses associated with programmers, engineering, QA etc. However, it is vastly more flexible to both vendor and consumer. It’s a great idea that has come into fruition. Great ideas are hard to stop.
References: [Note software programmers start at zero – binary]