The EvoStream Media Server can reduce your Total Cost per stream by 80% compared to the leading Java Based Streaming Server. There are many costs associated with operating a streaming service, but typically hardware, hardware maintenance and software licensing costs are chief among them.
The key EvoStream Media Server cost-saving driver is its efficiency which allows a server to handle 4 times as many concurrent streams. This means one server can now do the job of 4, drastically increasing your capacity or drastically reducing the amount of hardware needed to accomplish a target project. In addition, not all hardware is equally capable; perhaps you choose to serve your 30,000 clients with a cheap Intel i5 based server instead of upgrading to the latest Xeon based server.
For larger deployments, EvoStream provides advanced load balancing schemas and bandwidth reducing features. Using the EvoStream Media Server in a clustered environment only multiplies the cost savings.
In addition, the price of the EvoStream license does include all premium features (Transcoder, DRM, nDVR) and optional modules (Web based UI, Web-service integration framework) while other solutions require paid add-ons.
This tremendous value is achieved by an ultra efficient software architecture along with utilizing the native C/C++ programming language instead of running on a complex Java platform.
Here is an example of the cost associated with running a single EvoStream Media Server to its full capacity compared to the leading Java Based solutions.
Hardware maintenance and management of server infrastructure is also an important factor in your hosting costs. Reduction in infrastructure reduces the burden placed on your IT resources.
The above figures may seem bold, but they are reality when your business is thriving and you are maxing out your servers to their full capacity. If you are not yet maxing out your servers you know that having extra “head-room” is critical so that you can concentrate on building your customer base and not worry about how you are going to support the increase in traffic.
The bottom line for your business is straightforward. You can serve many more customers and lower your cost per stream dramatically, while delivering a high quality viewing experience.
Introduction
This white paper discusses the scalability test results and their implications on the Total Cost of Ownership of a streaming server solution. In these tests the EvoStream Media server is compared to a leading Java Based streaming server.
These test have been conducted by the EvoStream team and are not audited by a third party. That said, we are disclosing all the details and data so that you can arrive at your own opinion.
A Technology that is Dramatically Different
Not all media servers are created equal. The EvoStream Media Server is designed to the core with efficiency in mind. This started with the strategic decision to use the natively compiled C/C++ programming language instead of running on a complex Java platform.
As a natively compiled application, the EvoStream Media Server runs faster and makes the best use of your computing resources, whether it’s CPUs, Network Cards or RAM/Memory.
Utilizing our innovative core software architecture, we leverage multi-core CPU’s to the maximum by scaling self-contained compact streaming engines that maintain the most efficient software execution. This can simply not be achieved by threaded Java Based solutions.
That’s our DNA and how we are 400% more efficient than Java Based media servers.
There is no Java to complicate your life. You can chose any operating systems you like, and you can run it on your own server, or in the cloud, or a mix of both to serve 30,000 streams from a common Intel i5 Desktop computer.
If servers are increasing in capacity every year, so are videos streaming requirements. Video streaming is becoming ubiquitous and computing requirements are increasing with more content, higher definitions, and with more screens becoming the norm. Our technology vision will prove even more efficient as the industry will require more computing power to stream 24/7 HD content on multiple screens.
Test Environment
To arrive at our efficiency results, we ran a series of tests directly comparing the EvoStream Media Server with a leading Java Based solution. Here are all the details.
Hardware and Software Configuration
Ubuntu 12 64-bit Operating System.
IBM Desktop Clone.
Intel Core i5-2320 CPU – 4 Cores at 3.00 GHz.
8.00 GB Installed RAM.
1 Gbit/sec Ethernet card.
Metrics
CPU Utilization measured via the Top Linux utility.
Memory usage measured via the Top Linux utility.
Network Traffic measured via the SAR Linux utility.
Load Scenario
Starting with a single live RTSP source stream with no active clients.
Every 120 seconds (2 minutes), connect 500 new RTMP clients.
After 5,000 active connections, remove 500 RTMP clients every 120 seconds.
Test is over when connection count is back to 0.
Test Results
We will review and discuss results in terms of CPU utilization, network bandwidth efficiency, memory utilization, scalability and reliability. In each of the result charts, the X-coordinate is the time in seconds. The load is first increasing from 0 to 5000 connections, and then decreasing from 5000 to 0 connections. We have outlined key milestones with grey vertical lines at 2000, 4000, 5000 connections and then 4000 and 2000 connections.
CPU Utilization
Y-coordinate: CPU % utilization by both solutions.
This chart shows the core of the software’s efficiency. The less CPU we use to handle the load, and the more linear it’s increasing, the better it is. The net benefit is greater scalability and predictability.
EvoStream serves 5000 connections with fewer than 20% of CPU usage while the Java based solution is using 50% of the CPU before it crashes. EvoStream is far more efficient from the beginning, and the gap widens as the load increases.
On a real-life production system, you can deliver up to 30,000 connections from an I5 CPU before you saturate your CPU.
This scalability is a direct result of our natively running C++ technology and our IO-driven system architecture. These performance metrics simply cannot be achieved with a Java based platform.
Network bandwidth Efficiency
Y-coordinates: the actual Megabits per second of throughput, as opposed to a percentage of usage. In our test, the computer had a 1Gbps network card. The higher the number the better it is.
This test shows how the media server manages the most critical resource, network bandwidth.
EvoStream is capable of rapid delivery of content through the network, and the gap widens as load is increasing. EvoStream is also far more dynamic and reactive in moving the content through the network and quickly accepts connections and saturates the network card while the leading Java-Based server fails under heavy load and shuts down.
Network saturation is reached earlier with the EvoStream Media Server simply because more connections are accepted quicker. An important factor of any system is how it will behave in extreme situation, or in the “edge case”. EvoStream is designed to elegantly handle full loads and provides stable, reliable and, most importantly, predictable behavior when hardware resources have been maximized.
On real life production system, your network bandwidth is one of your most critical resources. The ability to maximize usage is critical to deliver your content to your viewers, and maximize any other resources from your server. There is no efficient CPU and memory use without an efficient Network management.
Ultra Efficient network bandwidth management is a core value of our software architecture. In addition, our compiled code is much quicker to operate with the network firmware.
Memory Utilization
Y-coordinate: % of memory actually used by the software.
This test is showing if and how the software is capable of efficiently utilizing the available memory. The test is two fold:
From 0 to 2,800 connections, EvoStream Media Server is using 66% less memory and increases its usage linearly, while the Java Based solution uses a much larger amount of memory with erratic usage.
After 2,800 connections, the network card bandwidth is used to its full capacity. EvoStream automatically buffers the streams in memory to reduce video packet-loss. This is why you see the memory usage scaling rapidly toward 50%. The Java Based solution does not provide any built in mechanism for preventing packet loss.
If memory utilization is critical to your system, this additional caching done by the EvoStream Media Server can be completely configured. It can easily be turned off or the maximum use per stream can be reduced, providing great control over the performance of your server as loads begin to saturate your network bandwidth.
On a real life system, EvoStream is way more efficient using much less memory. It is also capable of using the remaining memory to mitigate temporary network saturation, which result in a flawless experience for your viewers. You can choose to disable buffering to maintain extremely low memory usage.
Our compiled technology natively requires much less memory, and our core software architecture design and network layer management has built-in memory buffering.
Scalability and Reliability
Y-coordinate: total number of connections accepted by the streaming server.
This test shows how fast the media server accepts each subsequent load of connections.
EvoStream accepts connections faster and is capable of handling all 5,000 connections despite the saturated network. The Java Based solution accepts connections slower and crashes before ever reaching 5000 connections. From 2,500 connections and up, EvoStream accepts connections 5 to 10 seconds quicker on average.
On a real life system your viewers will be delivered their stream quicker, especially under serious load of your server. You will not lose viewers due to connections being rejected, or much worse, your server crashing due to overload.
Overall Results
EvoStream uses much less CPU and memory, releases resources much quicker, and handles heavy load even when the network bandwidth has been saturated.
The results of increasing load have been laid out above, but it is just as interesting to discuss the results when the load is decreasing. EvoStream shows similar resource usage when scaling down as when scaling up, demonstrating a balanced and symmetric load profile. This provides the stability and predictability over time necessary to run professional services
On a real-life system and when your business thrives, you may experience peak loads all the time. EvoStream can deliver and handle it whether it’s after 1 peak load or 100 because your server remains clean and ready to take on the next peak. EvoStream is Ultra Efficient, Scalable and Reliable.
TCO discussion
TCO is an approach popularized by the Gartner Group to quantify the financial impact of deploying an information technology product over its life cycle. In this white paper we narrow down the term TCO to the acquisition cost of a complete streaming server (software and hardware). We will now discuss those costs.
Cost of streaming license: This cost is significantly different for both solutions and their perpetual licenses that are unlimited in hours and sessions, users or streams. The price of the EvoStream perpetual license is $999. The Java Based leading solution perpetual license is priced at $1495.
Cost of add-ons and optional modules: The $999 price of the EvoStream perpetual license does include all premium features (Transcoder, Drm, nDvr) and optional modules (Web based UI, Web-service integration framework). The Java Based leading solution options (Transcoder, nDVR, DRM) are priced $495 each in addition to the server license, or offered in a promotional bundle priced at $1,695 when buying the server and the add-ons license all at the same time.
Cost of physical or virtual servers/computers: A more powerful server will deliver more streaming capacity while the cost of the streaming software licenses stays the same. With the EvoStream Media Server you can purchase cheaper, less capable hardware, and still deliver the same quality of service as the Java-based solution on expensive high-end servers.
Cost of technical support & maintenance: Both solutions offer online technical support and maintenance with their licenses at no extra cost.
Cost of training: Both solutions require people trained in streaming, operating systems and streaming server. The EvoStream solution relies on a simpler technology stack and will not required Java platform management skills.
CONCLUSION
The EvoStream Media Server can reduce your Total Cost per stream by 80% compared to the leading Java Based Streaming Server. There are many costs associated with operating a streaming service, but typically hardware, hardware maintenance and software licensing costs are chief among them.
The key EvoStream Media Server cost-saving driver is its efficiency which allows a server to handle 4 times as many concurrent streams. This means one server can now do the job of 4, drastically increasing your capacity or drastically reducing the amount of hardware needed to accomplish a target project. In addition, not all hardware is equally capable; perhaps you choose to serve your 30,000 clients with a cheap Intel i5 based server instead of upgrading to the latest Xeon based server.
For larger deployments, EvoStream provides advanced load balancing schemas and bandwidth reducing features. Using the EvoStream Media Server in a clustered environment only multiplies the cost savings.
In addition, the price of the EvoStream license does include all premium features (Transcoder, DRM, nDVR) and optional modules (Web based UI, Web-service integration framework) while other solutions require paid add-ons.
This tremendous value is achieved by an ultra efficient software architecture along with utilizing the native C/C++ programming language instead of running on a complex Java platform.
The bottom line for your business is straightforward. You can serve many more customers and lower your cost per stream dramatically, while delivering a high quality viewing experience.
Any questions? Please get in touch!
Thank you for choosing EvoStream.
The EvoStream Team.
EvoStream is a registered trademark of EvoStream Corporation. Various product and service names referenced herein may be trademarks of EvoStream Corporation. All other product and service names mentioned may be trademarks of their respective owners.
Copyright © 2015 EvoStream Corporation all rights reserved.
