Turbonomic (company)

Turbonomic provides an Application Resource Management (ARM) and Network Performance Management (NPM) platform. It connects business-critical applications to the underlying supply chain of resources to monitor and manage resource relationships and risks to performance throughout all layers of the network, including applications, containers, cloud, on-premises software, and more.

The platform adopts an application-aware approach that leverages insights into applications’ fluctuating demand and configuration to dynamically match demand to the required resources. Turbonomic promises anywhere from 15-65% savings per action, by deleting waste, scaling storage and computation, suspending workloads, and leveraging cloud discount models. It features:

• Insight into the complete application stack, software and hardware, enabling users to see how the top business applications and top business transactions are performing in the context of the application stack.

• Insight into how every entity consumes resources, allowing the user to navigate to the applications seeing the most significant increases in demand and to locate performance risks.

• APM, Kubernetes, cloud, hybrid, and multicloud support.

An overview of the Turbonomic platform

Overprovisioning and always-on resources contribute to wasted cloud spend that is expected to exceed $26.6 billion in 2021. This is caused by a lack of visibility, increasingly complex multi-cloud environments, and inaction on the part of business units in regards to addressing the potential for cost reduction.

Turbonomic uses multiple metrics and cloud constraints in its analysis. It generates prescriptive actions based on real-time metrics and IT policies that can improve application performance and reduce the workload of application teams. These actions are fully automatable and can be integrated with ITSM (IT service management) or approval workflows.

Shmuel Kliger co-founded Turbonomic with Danilo Florissi, Yechiam Yemini, Shai Benjamin, and Yuri Rabover. Shmuel Kliger was president of the company for its first twelve years, heading product development, strategy, and management. Before founding Turbonomic, Shmuel served as Vice President of Architecture and Applied Research at EMC's CTO Office (now part of Dell Technologies). Prior to that, he was the CTO of the Resource Management Software Group at EMC, which he joined through EMC's acquisition of Smarts where he was a co-founder and CTO.

Turbonomic was launched in 2009 and is headquartered in Boston, Massachusetts. Its subsidiaries include SevOne and ParkMyCloud. Together with its strategic partners, including Cisco, IBM, Microsoft, and AWS, Turbonomic serves more than 3,300 customers, including 36% of Fortune 500 companies. Turbonomic was formerly known as VMTurbo, before being renamed in 2016.

Turbonomic’s research found that more than 60% of all VMs (virtual machines) are oversized. The company’s resizing solution, which adjusts the size of the VM to match particular resourcing needs, can result in a cost reduction of 30-40%. In contrast to other scaling solutions that estimate data size in accordance with performance peaks or averages, which may result in mismatched capacity, overspending, or performance failures, Turbonomic uses percentile-based scaling to achieve cloud elasticity and provide greater accuracy and stability.

Turbonomic cloud volumes optimization involves mostly non-disruptive actions that increase volume sizes to improve performance (IOPS and throughput) and modify the provisioned capacity of advanced volume types (Azure Ultra, EBS IO1, IO2, and GP3).

Turbonomic provides full visibility of PaaS services and optimizes database PaaS services (Azure SQL and Amazon RDS) as well as container-based PaaS services (AKS, EKS, and GKE). Optimization actions include database scaling, containers scaling, and continuous pods placement. These actions result in near-zero downtime impact. Customers that employ optimization on PaaS resources yield up to 66% additional savings on top of savings gained from Infrastructure-as-a-Service (IaaS).

Turbonomic provides specific and automatable RI-aware compute scaling actions to increase existing RI (reserved instance) inventory utilization. Turbonomic RI purchasing actions, which are based on observed utilization, maximize reservation-to-VM coverage to minimize costs. Turbonomic automatically ingests and displays negotiated discount rates as part of an AWS EDP or Microsoft Azure Enterprise Agreement.

Since non-production cloud resources (such as those needed for development, testing, staging, and QA) are typically only needed during the workday, they should be deactivated outside of operating hours. Non-optimized systems keep these resources active with on average 65% more uptime than necessary, significantly contributing to the overprovisioning problem.

Turbonomic’s ParkMyCloud platform uses resource utilization history to automatically recommend on/off schedules for resources to optimize costs. ParkMyCloud manages compute, scaling groups, databases, and storage on all major public clouds, as well as nodes within AKS, EKS, and GKE clusters.

Turbonomic’s cloud migration planning and modeling engine aids users in the optimization of cloud migration by using application-aware historical utilization data to select the best VM/instance type for every resource and RI recommendation for steady-state workloads.

The insight into the application and infrastructure stack provided by the platform can be used to determine whether issues are caused by application code or application resource contention. The automatization of actions can completely prevent resource contention.

While Kubernetes provides a platform to manage and orchestrate containerized environments, unlike the Turbonomic platform it does not make application resource decisions. Additionally, Turbonomic offers the following features:

• The software sets service level objectives (SLOs) and automatically ensure they are continuously met.

• The platform can be set to full automation and execute actions in real time, or the user can execute them as part of the existing processes, e.g. CI/CD (continuous integration/continuous delivery).

• It visualizes and informs how services correlate to the platform and the underlying infrastructure and its analytics continuously make optimal resource decisions. For instance, it will not determine that scaling out pods is necessary without ensuring that there is enough capacity in the nodes to handle the workload.
• The platform scales container limits/requests up or down based on application demand, freeing developers to focus on application features and functionality.
• Turbonomic reschedules pods while maintaining service availability to avoid resource fragmentation and/or contention on the node, which increases density while ensuring that pods can always be deployed.

• The software models prospective scenarios based on the real-time environment by determining how much surplus space there is in the clusters or simulating adding or removing demand (Kubernetes pods), enabling planning the onboarding of new services.

IBM acquired Turbonomic on April 28th, 2021. IBM plans to leverage Turbonomic’s Application Resource Management (ARM) and Network Performance Management (NPM) capabilities. Following the acquisition, IBM is the only company that can provide customers with AI-powered automation capabilities that span from AIOps (the use of AI to automate IT Operations) to application and infrastructure observability. Red Hat OpenShift, on which the system is built, will allow it to operate in any hybrid cloud environment.