Turbonomic’s AI approach seeks to unify the full application stack
In an application-driven enterprise world, software should manage IT.
This is the founding principle behind Turbonomic Inc. and its belief that software can assure optimal application performance in real time. It takes artificial intelligence and machine learning to allow applications to smarten up resourcing decisions in a complicated, highly monitored stack.
“The world is over-alerted,” said Ben Nye, chief executive officer of Turbonomic, during the “Apps ON Cloud Summit” in May. “What we really want to do is to make these applications pick the resources on which they want to run to assure their own performance. The role and the opportunity for AI are to help assure the application is performant, maintain its policies and compliance, run the app, and at the same time deliver a much more cost-effective experience.”
Supporting the full stack
The role of AI is a key element in the evolution of application resource technology. It is responsible for replatforming applications away from legacy virtualization, and creating a new paradigm where a highly performant network can be delivered at scale.
To bring this new reality to life, Turbonomic set out to address the tiering issue. In today’s IT stack, many tiers need to be optimized, including compute, network and storage; virtualized containers; pods; and a scheduler.
“The more tiers you have, the more tears you’ll have, because the performance on the application at the top is so removed from the resourcing decisions all the way down through the stack,” Nye explained, claiming Turbonomic is. “the only solution that unifies what we call the ‘full application stack.’”
Applications need resources to run effectively. By ingesting technology into the physical and logical resourcing tiers, normalizing it and then leveraging AI so applications have the freedom to choose what is needed in a common data model, customers can deliver services more reliably.
“We took all of the resources on which the application performance depends and then served them up in a marketplace,” Nye said. “Without that, resources congest and it starves the application’s performance. You see that loading, spinning wheel of death, which is brutal to a business brand or its return on innovation. Eliminating that was really the key here.”
Shift to action
Turbonomic’s approach to application performance highlights an important byproduct of AIOps and its growing role in enterprise IT. AI is enabling a shift from systems of record to systems of action.
The company’s desired state is to turn data into actions that continuously assure application performance. APIs discover platforms and infrastructure, analytics process the data, software makes the decisions and action is the result.
System monitoring tools gather in mountains of data, far more than a human can process for decisive action. With the proper policies, AI can assess that recorded data at scale and take action.
Turbonomic’s acquisition of SevOne Inc. in 2019 offers an instructive look at how this scenario plays out in the modern enterprise. SevOne provides network performance and monitoring diagnostics based on a proactive analytic approach that understands system context.
“The challenge is that networks are ‘bursty’; sometimes traffic is high and sometimes it’s not,” said Tim Greenside, senior solutions architect at SevOne, during an “Apps ON Cloud” session. “Why not learn what normal is and put yourself in a better position so you don’t have to try and make a good guess for a reasonable threshold? You can use those baselines to learn what normal is and then have automatically adjusting thresholds based upon what normal is and standard deviation.”
Scaling for Kubernetes
Turbonomic has also embraced automation in the container space. The firm offers automated container resizing through its Fast and Accurate Container Planning capability.
IT organizations can scale resources for Kubernetes clusters using real-time automation in a cloud native application. Turbonomic has applied this technology to support a client in the travel insurance carrier business with ties to one of the country’s largest low-cost airlines. During a busy Easter holiday travel period, Turbonomic’s dynamic resourcing on the Kubernetes platform reportedly maintained a key application’s performance high for its insurance customer and the response time low.
The use of AI to manage application resources also has a direct impact on cost. A recent StormForge survey of IT professionals in North America found that almost half of cloud resources spending was being wasted. Reducing that waste was a priority for 76% of the respondents.
The key is to move from an allocation model to one tied to consumption, which Turbonomic’s offering is geared for, according to Nye.
“Allocation is a fancy word for guess,” Nye said. “But with consumption you can know exactly how much resources, exactly the order, for exactly the point in time. A guess can never be as exact as consumption. That’s what we’ve really built.”
Best practices for efficiency
Turbonomic offers a number of best practices to consider when managing the full application stack.
Because today’s IT environments are in a constant state of flux, it’s generally a good idea to revisit the state of applications and infrastructure when new software is installed or incremental updates are made.
It’s also good practice to adhere to principles of cost optimization when operating in the cloud. This involves understanding what resources applications are consuming across the stack and ensuring that performance remains as cost efficient as possible.
The process of training machine learning models can also leverage operational efficiency based on what SevOne’s Greenside refers to as “business hours.” In a retail organization, for example, the prime hours are generally from 8 a.m. to 10 p.m. Off hours place less demand on the system, which becomes important when allocating resources for peak application and network performance.
“Your machine learning modeling is only as good as the data you are getting,” Greenside noted. “By having ‘business hours,’ you can ignore those down times because you don’t want to water down your averages. The system will learn what’s normal. We want to plan and make sure we have adequate capacity during the noisiest part of the day.”
In April, IBM announced it would acquire Turbonomic for a reported $1.5 billion. The move was viewed as another chapter in IBM’s ongoing effort to strengthen its focus on high-growth cloud and AI markets. The acquisition also reinforced the continued importance of applications and automation in the IT world.
“Getting insight and visibility into what the business is doing and why it matters helps people make more informed decisions,” said Corey Quinn, chief cloud economist at The Duckbill Group, during an “Apps ON Cloud Summit” interview. “The more along those lines you’re able to automate, the better the outcomes are going to be.”
(Disclosure: TheCUBE is a paid media partner for the Apps ON Cloud Summit event. Neither Turbonomic Inc., the sponsor for theCUBE’s event coverage, nor other sponsors have editorial control over content on theCUBE or SiliconANGLE.)
Image: Everything Possible
A message from John Furrier, co-founder of SiliconANGLE:
Show your support for our mission by joining our Cube Club and Cube Event Community of experts. Join the community that includes Amazon Web Services and Amazon.com CEO Andy Jassy, Dell Technologies founder and CEO Michael Dell, Intel CEO Pat Gelsinger and many more luminaries and experts.
We really want to hear from you, and we’re looking forward to seeing you at the event and in theCUBE Club.