Concertio taps AI to optimize apps and servers in real time
AI-powered optimization software provider Concertio today announced that it has raised $4.2 million, which it says it will use to scale its platform that boosts app performance by tailoring x86 server settings. Customers won’t only benefit from faster, more stable apps and hardware as a result, says Concertio, but from lower infrastructure costs as well.
Concertio’s Optimizer Runtime tool, which runs on most CentOS and Debian Linux distributions running in the cloud or on-premises, monitors and learns from the interactions between apps and systems to tune parameters at each stage of development and deployment. An agent runs in the background and learns system metrics to detect different phases of execution, ingesting only a small sampling of data while dealing with system variability to optimize each phase separately. The Optimizer Runtime discovers hidden systems configurations that deliver optimal settings, and it optionally lets developers define optimization targets, including processor instructions, network bandwidth, database queries, and minimum server energy.
Concertio’s Optimizer Studio — which in contrast to Runtime runs on a range of hardware, including networking cards, CPUs, ASICs, FPGAs, storage appliances, mobile devices, and complete systems — works in tandem with Runtime to make apps and systems operate in concert for improved performance. Studio supports many tunables out of the box (such as task affinity, NUMA page migration, choice of IO schedulers, task scheduling granularity, DVFS policy, symmetric multithreading, and CPU last level cache prefetching) and allows developers to add custom tunables with scripts if they aren’t supported. Studio also helps those developers discover market-ready default configurations for hardware products, and it can enable support teams to provide extended support to customers, chiefly by reproducing performance issues in virtualized settings.
Concertio’s suite offers three modes of optimization: agent-based dynamic optimization for use in production servers, continuous optimization where static optimization is implemented within the continuous integration/continuous delivery pipeline, and static optimization for use by performance engineers and IT professionals. The company says that on average, its customers — which include Intel, Mellanox, and Marvel — see above 10% improvement in performance over baseline untuned systems using Studio alone and up to an 8 times improvement in certain use cases.
“We’re entering the era of self-tuning servers, where servers automatically adjust their settings dynamically in real time according to the workloads that they run,” said Concertio cofounder and CEO Tomer Morad. “Our Optimizer products transform general-purpose systems into high-performant special-purpose systems, thereby boosting performance and slashing infrastructure costs.”
Differential Ventures led New York-based Concertio’s latest funding round.
Source: Read Full Article