We are constantly espousing the technical advantages of the Data Distribution Service (DDS) standard. It is truly far more technically capable than alternative middleware systems such as AMQP, JBOSS, MQTT and internally-developed efforts. However, we don’t spend enough time discussing the financial advantages of DDS solutions.Looming budget cuts (i.e., sequestration) hang heavily over government and business entities. We don't have to consider the situation all doom-and-gloom, though. It may be the perfect opportunity to adopt better and more affordable solutions. Both governments and businesses can embrace options for superior technology that costs less to develop, implement and maintain.
Let’s take a concrete example.
The Army’s Joint Battle Command-Platform (JBC-P) system set out to track positions of friendly and hostile forces on the battlefield. Their goal was 100,000 tracked updates per second (tracks). After spending 8 years developing 500,000 source lines of code (SLOC), they needed 21 servers to deploy and operate their Network Operations Center (NOC). In the end, they only managed 20,000 tracks at peak load, with significant reliability and uptime issues.
Now let’s look at the financial impact.
We can estimate the cost of development and somewhat predict the overall maintenance of this system. Average SLOC costs range from $15-25 per line. This probably doesn’t include rigorous testing and certification costs. Let's be extremely conservative, and use $20 per line.
500,000 SLOC x $20/line = $10 million for development
Just to look at things another way, say they had 20 developers at $120 thousand per year per developer.
20 developers x $120,000/developer/year x 8 years = $19.2 million in coding expertise over 8 years
Software lives on hardware. We can estimate the hardware and software costs of a server as $10,000.
26 systems x $10,000/system = $260,000 in hardware
Lastly, we need to consider ongoing costs for maintenance and support. Supporting a system of this size might take 30 hardware and software engineers.
30 engineers x $120,000/engineer/year = $3.6 million in maintenance and support personnel per year
Our conservative grand total for development comes out to $10-19.2 million (depending on whether we use SLOC or developer costs), with an ongoing support cost of $3.6 million/year.
And in the end, the system didn’t perform to spec, couldn’t scale, and wasn’t reliable.
Of course, that wasn’t really the end. This is the U.S. Army, after all. They wanted to get it right. Their next-gen effort reviewed several middleware solutions. After several months of evaluating capabilities, scalability and reliability, DDS was ultimately chosen to fill their technology need.
It sure helped that they were able to build a working DDS-based prototype in under a week. RTI was proud to contribute a critical architecture study and training programs to the project.
The new system delivered eye-popping results:
|SLOC||$20/line||500 K lines||$10,000,000||88 K lines||$1,760,000||82%|
Did I mention that the DDS solution “broke” their testing framework by supporting 250,000 tracks with full redundancy? They expected that the DDS-supported system would handle 500,000 tracks easily – 5 times the specified requirement.
By the way, these savings are per NOC. The Army had 5 NOCs: 3 for production and 2 more for development, training and backup.
These are typical results. Whether you choose to base costs on a per developer or SLOC model, DDS based-solutions present a compelling alternative to existing middleware implementations.
Even without impending sequestration, DDS creates compelling business opportunities. Every company, as it evolves, is seeking out more efficient and economical solutions to their needs.
The bottom line:
For about 25% of your existing annual system maintenance costs, you can develop and deploy an enduring DDS-based solution that delivers return on investment (ROI) within the first year of implementation.
How much money would you save with a DDS solution?