We’re excited to bring Transform 2022 back in person on July 19 and pretty much July 20-28. Join AI and data leaders for insightful conversations and exciting networking opportunities. Register today!
When Hakim Weatherspoon looks at a cloud, he sees inefficiency everywhere. Sometimes the instances are overprovisioned and the extra RAM and CPU power costs money even if it is not being used. Sometimes the code runs on an overpriced machine. The cloud companies make it easy to turn on the power when the demand is high, but it’s not so easy to cut it.
So he builds tools to squeeze these inefficiencies out of the cloud. He is the CEO and co-founder of Exotanium, an Ithaca, NY startup building a software layer that makes it easier to run code and migrate to the cheapest hardware. It juggles containerized code by constantly optimizing performance and growing and shrinking its dedicated hardware in response to demand. It is designed to work in all clouds.
“The people who are in the most pain are the ones who spend a ton on computers,” Weatherspoon says. “People who spend a lot on simulations or other forms of high-performance computing because they don’t have a good way to reduce their computational load.”
Reduce calculation costs
This week, the company released the results of a study it completed with the Idaho National Laboratory. The project used Exotanium’s xSpot tool to orchestrate some of the lab’s computational experiments, sometimes lasting hours at a time. In the end, the bill for the calculation fell by more than 70%. The test examined the performance of MASTODON, a “high-performance computing simulation for structural dynamics, seismic analysis and risk assessment” packaged in a Docker container and running on AWS.
The savings came from running the project using machines rented from Amazon’s spot marketplace. All clouds are constantly auctioning their backup compute cycles, and in times of low demand the savings can be 90% or more.
“The spot market is cheap, but not reliable,” Weatherspoon says. “We are able to take the application as it is and migrate without stopping it on the spot market and then leaving again [to standard instances.] It’s kind of a whack-a-mole game without letting me whack.”
Exotanium has created a low-level virtualization technology that simplifies migrations. xSpot takes into account factors such as the current spot price for computers to predict when the cloud may shut down an instance. When it looks like a spotting machine shutdown is imminent, it moves the process and all its state to a more expensive one until the market opens.
“We have proprietary algorithms that will put in price and capacity and rebalance signals and a lot of other things, and we’ll kick off the migration well in advance,” Weatherspoon said. “We succeeded so well that we have been on site for months, actually six months. Without being terminated.”
The company also uses the same low-level techniques to repackage running code into hardware in the most efficient way. xStack will consolidate running machines so that the minimum number of instances is used. xScale dynamically adjusts the size and power of the instances to meet current demand.
In some cases, the Exotanium layer can simplify software running by removing extra layers that may not be necessary for long-running code.
Puts you in control in the cloud
“In this Idaho National Lab case, we’re faster than their original implementation,” Weatherspoon says. “The virtual machine’s hypervisors provide isolation. We then run the operating system at the same privilege level as the application. What that effect is is that now all system calls – every call from the application to the operating system – are a hundred times cheaper. †
The technology is designed to work with other popular container management options such as Kubernetes, giving devops teams more options and control over cloud costs.
“If you go to the cloud, you don’t have [virtualization layers like hypervisors] not because Amazon or Google or Microsoft own that physical machine and they own that layer of software that virtualizes that machine that you as a user don’t have access to,” explains Weatherspoon. “But our product does. It gives you that access back.”
Exotanium has offices in Ithaca, New York and Santa Clara, California. Nearly a year ago, Exotanium announced a $5 million seed funding round led by Walden International and Nepenthe Capital LLC.
VentureBeat’s mission is to be a digital city square for tech decision makers to learn about transformative business technology and transactions. Learn more about membership.
This post Exotanium aims to optimize cloud efficiency (and reduce costs)
was original published at “https://venturebeat.com/2022/04/08/exotanium-releases-new-tool-to-reduce-computational-load/”