Sustainability at Zixi

Zixi corporate & social responsibilities include investing in more efficient processes that can actively help mitigate Zixi’s carbon footprint.  This post attempts to quantify those efforts and set some targets for 2022.

Summary

Video delivered using Zixi is significantly more efficient than other means of networked video delivery.  The process of producing and delivering live video has many areas where impactful efficiencies can be introduced, the following are specific to innovation at Zixi that are producing quantifiable results:

  • Bandwidth Utilization (20%+ efficient bandwidth utilization than other protocols)
    • Content aware trans-rating
    • Dynamic intelligent error correction and packet recovery
  • Video Traffic Throughput (1.5x to over 6x greater data throughput)
    • Zixi processor and memory utilization optimizations deliver market leading throughput
    • Zixi v15 release (Q1’22) doubled market leading throughput
  • Compute Resources (up to 40%+ more efficient compute processing per unit of consumption)
    • ARM processor support
    • Compute dense transcode processors

Comparison Criteria

In this post, we have taken synthesized customer data and extrapolated out estimated greenhouse gas (GHG) emissions leveraging three different implementation approaches.

  • Legacy Transport Protocol:  Estimates efficiency when delivering video using standard RTP w/FEC
  • Zixi Protocol v14:  Estimates efficiency when delivering video using Zixi Protocol v14
  • Zixi Protocol v15+:  Estimates efficiency when delivering video using Zixi Protocol v15+ (v15 available Q4’21 / v16 available 1H’22)

While this example uses real world data to estimate volumes, there are necessary assumptions on “perfect world” utilization that are likely not achieved in production environments.  Even with that, the estimates provide verifiable results that help define the scale and direction of delivery efficiency and sustainability.  Zixi will be periodically updating and enhancing the data in this sustainability report as new information and modeling approaches are implemented.

Customer Snapshot

In order to produce this report we took a look at the video delivery data across several Zixi ZEN Master users.  We then averaged out the customer data to accurately model a typical operating environment.  The results showed that the customer average of approx. 76PB of live video delivered, through 2500 video flows at an average of 7.4Mbps per flow and across 198 Zixi instances.  A video flow typically consists of one or more sources trafficked to one or more target destinations.

  • Annual Volume: 75PB of Video Delivered
  • Avg Live Video Bitrate: 7.65Mbps
  • Total Flows (Sources + Targets): 2500
  • Avg Active Servers: 198

With this data, we can work to estimate GHG emissions associated with this infrastructure.  We can also model estimates that can highlight how sustained investment in innovation is producing enhanced efficiency that directly impacts sustainability priorities.

Bandwidth Utilization

The Zixi protocol continuously and automatically optimizes bandwidth use based on changing network congestion and contextual changes in the video.  The protocol also manages error correction dynamically and is the only protocol to support both push and pull protected multicast strategies that can further reduce overhead associated with video distribution.  The Zixi Software Defined Video Platform flexibly supports all modern video delivery protocols, including RTMP, RIST, SRT, HLS and more.  However, due to the improved performance, reliability and efficiency, the majority of Zixi ZEN Master managed flows are using Zixi protocol.

In practice, the Zixi protocol achieves approx 20% bandwidth efficiency per stream as compared to legacy RTP w/FEC.  This is attributed to significantly less overhead associated with ensuring sustained reliable delivery.

Bandwidth savings can be quite impactful with regards to both costs and sustainability goals.  According to a recent study, it is estimated that 0.044kWh of electricity must be consumed to deliver 1GB of data over US based networks (2021 estimate).  The Estimated US Energy Mix emissions rate for 2021 is 0.39kg CO2 equivalent for every kWh of electricity.  Using these values, we can begin estimating what the carbon footprint associated with live video distribution was last year using the different comparison models.

  • Network Utilization: 75,000,000GB
  • RTP w/FEC: 3,300,000 kWh consumed | 1,288,000kg CO2 equivalent
  • Zixi v14/v16: 2,600,000 kWh consumed | 1,030,000kg CO2 equivalent
  • GHG Emissions Reduction (annual): 257,000kg CO2 equivalent

Improving delivery efficiency by 20% using the Zixi protocol produces GHG emissions savings estimated at 257,000kg CO2.  As more customers and partners leverage Zixi for and through continued investment in research and development yielding improved efficiency rates, further improvements will be achieved.

Video Traffic Throughput

ZEN Master managed video flows often consist of multiple sources delivered to multiple target destinations.  In ZEN Master, these flows are managed as a channel.  A single channel will often have a pair of bonded sources supporting hitless failover resilient distribution delivered to internal production and external distribution partners.  ZEN Master also monitors the video objects and automates processing tasks such as PID mapping, packaging, protocol handling, and more.

Throughput measures the volume of concurrent video flows that can be handled in a single compute instance.  At Zixi, we measure this by calculating the maximum bandwidth per processor core.  These measurements are highly variable, with dependencies on workload composition and processor core performance.  At Zixi we do perform internal testing to understand capacity and work with customers to optimize their throughput based on these variables.

Measuring gains in throughput efficiency is platform dependent, so for this exercise we will be estimating efficiency gains within the Zixi SDVP.  Throughout 2021, Zixi made significant engineering investment improving overall throughput.  These improvements have resulted in throughput efficiency gains measured between 33% and 100% (double) operational throughput.  While there is no absolute value given the variabilities mentioned above, we can reasonably assign a 50% throughput improvement value between Zixi v14 and v15.

The synthetic customer data used for this testing indicates a baseline of 198 servers yielding an annual throughput of 75PB of delivered data.  This factors out to each server effectively managing approx. 100Mbps (12.5MB/s) of video assets.

  • Baseline Server Throughput: 12.5MB/s
  • Baseline Server Count: 198
  • V15 Server Throughput (estimated +50% efficiency): 18MB/s
  • V15 Server Count Req’d: 132

An avg compute instance running Zixi Broadcaster will have 8 CPU cores.  Research in to instance GHG emissions rates for cloud compute instances yields a compute optimized 8-core instance producing 32.3gCO2/hr when operating in US based data centers.  Estimating the GHG emissions per server instance at 32.3g CO2/hr:

  • 198 Compute Instances (v14): 56,000kg CO2/yr
  • 132 Compute Instances (v15): 35,600kg CO2/yr
  • GHG Emissions Reduction (annual): 17,800kg CO2 equivalent

For our synthesized customer, simply updating to the latest Zixi releases yields 17,800kg CO2 equivalent reduction in GHG emissions and likely significant cost savings by lowering the total number of deployed instances required in the production operating environment.  Zixi expects that number to continue to grow as additional processes are optimized.

Compute Resources

In 2021 Zixi began implementing support for new ARM based cloud instances.  With the latest Zixi updates, customers can now begin leveraging ARM compute instances in the cloud.  Not only do these instances have comparable or improved performance over similarly spec’d x86 instances, they offer both cost and emissions savings.  By consuming less power at the same level of operating performance, ARM processors can eliminate significant emissions associated with the compute layer.  Zixi expects to offer ARM processor support in 1H’22 with Zixi v16 updates.  While full testing is required to quantify results, and as with throughput there is significant variability in the test scenarios that will yield different results, initial emissions estimates provide GHG emissions rates:

  • 8-Core Compute Instance (x86): 30.8gCO2/yr
  • 8-Core Compute Instance (ARM): 16.3gCO2/yr

If our example customer were to fully implement the same number of ARM based instances as they are running today, their carbon footprint will be significantly lower:

  • 132 Compute Instances (v15 on x86): 35,600kg CO2/yr
  • 132 Compute Instances (v16 on ARM): 18,800kg CO2/yr
  • GHG Emissions Reduction (annual): 16,800kg CO2/yr

The best part here is that the comparable compute ARM instances are generally less expensive than equivalent x86 instances, making greener infrastructure work for the environment and the budget.

Conclusion

In total, using the synthesized data to estimate a real world customer operating environment, this customer emissions rates would be as follows:

  • Baseline: 1.341M kg CO2 equivalent
    • Network: 1.288M kg CO2 equivalent
    • Throughput & Compute: 53.4k kg CO2 equivalent
  • Optimized: 1.047M kg CO2 equivalent
    • Network: 1.030M kg CO2 equivalent
    • Throughput & Compute: 18.8k kg CO2 equivalent
  • GHG Emissions Reduced: ~295,000kg CO2 equivalent PER YEAR

According to the EPA GHG Equivalence calculator, 295,000kg CO2 per year equates to:

  • 741,000 fewer passenger miles driven
  • 683 fewer barrels of oil consumed
  • 4878 tree seedlings grown for 10 years

At Zixi, enabling our customers to exceed their operational goals in the most eco-friendly way is a core principle that informs every decision we make.  Throughout 2022 and beyond, we will be updating our data, improving our estimates and contributing to the body of knowledge on how our work and innovation impacts our ecosystem.

Learn more about what Zixi is doing to power next generation globally distributed linear OTT services today.

Zixi enabled universal gateway, automating ingest and normalization of hundreds of linear OTT channels for leading video services.

Zixi guarantees compliant streams to any destination with on-the-fly digital supply chain processing.

Zixi ZEN Master Control Plane enables remote production and management with no compromises to quality or visibility.

Zixi’s commitments to deliver premium quality services, innovations to support sustainability efforts and a culture of inclusiveness that values diversity!