“Where we have implemented Opsani, the teams have been able to get 50-100% more out of their resources.”
Ancestry oversees an online genealogy network of over 15-million people and more than 10-billion historical records. With tens of thousands of components processing petabytes of data, there is a lot of room for underutilization and wastage.
Fortunately, Ancestry has leveraged the capabilities of Opsani’s AI platform to optimize their cloud operations. In an interview with Darek Gajewski, Principal Infrastructure Analyst at Ancestry, we went into detail on how Opsani has helped Ancestry, and why other companies should also take steps to begin the Continuous Optimization of their systems.
Darek described how Ancestry have gradually been implementing Opsani across all of their various systems. They have been onboarding their service teams and letting them pick at least one component from their 40 major services on which to try Opsani:
“They’ll usually go, ‘Okay, let’s start with a backend component, something that might not be customer-facing…’ And then they’ll implement with larger stats and components.”
The onboarding process, Darek explained, gives service teams the ability to pick the component they want to integrate Opsani into. Once they start to feel familiar and comfortable with the tool, they can start adding other components. Which they do, fast, because the results are compelling:
“Where we have implemented Opsani, the teams have been able to get 50-100% more out of their resources… Once implemented, the tool does the optimization, and it frees the team to do feature development and other aspects of their job. That’s really what’s vital to Ancestry.”
From the first POC that Opsani did with Ancestry, the results have shown a 51.7% reduction in cost with no performance degradation. And as Opsani continues to optimize Ancestry’s systems, this value can only get better.
Darek discussed how such an AI tool could be significant to other companies, especially for those who aim to grow and scale while making sure that the systems still function correctly:
“At small scales, we don’t typically worry about our costs, but then when you get to $1M, $10M, or $20M in costs, you find that, ‘Oh, these systems are really expensive all of a sudden.’ And then we’re forced to go back in time to reengineer or re-architect something that’s very large and expensive.”
At the current state of things, companies are gearing up to move their infrastructures and systems into the cloud, if they haven’t already. There are two ways to do this: the “lift-and-shift” approach, and the “rebuild the system in the cloud” approach.
Darek weighs in on both, and points out the inefficiencies of each one, from sticker shock with regards to system costs to valuable time wasted on re-architecting data systems to be cloud-native.
“That’s what Opsani’s helping us with. If you can implement this, it can save you time. The continuous portion of that optimization is important because the entropy of the system, if not maintained, increases over time. But if you can continue to watch that system, you can stem the entropy a little bit and make sure that your inefficiencies are not coming back.”
Talking about the future of AI systems, Darek envisions how Opsani could further evolve to not just optimize individual components, but deal with interconnected systems as a whole and stem problems like cascading failures.
“If you can watch those and understand your dependencies and the scaling factors of your entire infrastructure, a neural net [like Opsani] should be able to leverage those moments and lift the entire infrastructure up for a little while to get over whatever hump you’re having.”
When asked about the goal that Opsani is helping them achieve, Darek says:
“Our goal, by the end of 2020, is to save over $1M in infrastructure with Opsani on 20 of our 800 stacks.”
With how Opsani is currently optimizing Ancestry’s multiple technologies, that goal might not be so far away. With Opsani’s assistance in ensuring maximum efficiency for their systems, there’s no telling how many more milestones this company will achieve.