We’ll pass on the obvious reference to “The Princess Bride'” but Electric Cloud is faced with a dilemma. We have incontrovertible proof that we save our customers millions of hours of developer time each and every month – based on anonymous ElectricAccelerator telemetry sent from customer sites. Yet the numbers are so high many people find them hard to believe.
For instance, in the month of July alone, just ten of our customers ran 115,000 builds with ElectricAccelerator consuming 1.5 million hours of CPU time and just over 37,000 hours in elapsed time, saving a total of 1.435 MILLION hours of developer time.
In one month.
And that’s just from a few of our customers.
And they are saving that time running their builds and tests on less than half the hardware they used to use.
Maybe they were slacking off in July because in June those same customers saved 1,489,000 hours.
What’s the Value?
How does that translate into value? If a typical fully-burdened developer costs $100 per hour (actual results may vary based on region, benefits packages, etc.), that’s nearly $150 MILLION dollars taken out of the cost of getting products to market in a single month.
Developers at these companies typically ended their day by kicking off a build and test run before they went home, hoping it will complete successfully. In the morning they would slog through logs to find out if their code change worked and if it passed all the tests. Now, developers are only waiting 15 minutes or so for their builds to complete so they can run multiple iterations in the day (and spend less time on the ping pong table).
Instead of being caught in a never-ending cycle of throwing ever more expensive hardware at the build and test speed problem, infrastructure managers are getting maximum utilization out of available hardware, often on 75% fewer cores. Now, they are seeing longer server life and reduced heating and cooling costs (or AWS bills) – and more pleasant budgetary meetings.
Instead of waiting weeks to test one change on 35 different devices and multiple versions per device, engineering managers know the same day if a change impacts delivery dates. Now, they get products to market faster (17 releases in a year instead of 4 for one company) and happier customers.
At the end of the day, these numbers are staggering but real.
Maybe if we cut the numbers in half people would be more likely to believe us…
If you don’t believe us, you can always try it for yourself here.
Latest posts by Tim Johnson (see all)
- AI, DevOps, and Employment Law: Thinking Ahead - December 10, 2018
- Loving Your Pipeline’s Middle Child: 60X Test Acceleration - December 7, 2018
- More Speed with ElectricAccelerator – Part 2 - November 28, 2018