Το Alluxio για διοχέτευση δεδομένων σε εφαρμογές σε υβριδικές πλατφόρμες cloud με 50 εκατομμύρια $

San Mateo, με έδρα την Καλιφόρνια Alluxio, a data orchestration platform, today announced that it raised $50 million in an oversubscribed series C funding round led by an unnamed backer with participation from a16z, Seven Seas Partners, and Volcanics Venture. Leveraging the new tranche, Alluxio plans to expand its presence in the Asia-Pacific region with the opening of an office in Beijing while bolstering its product development, go-to-market, and engineering operations.

A growing number of companies keep their data on-premises while running apps in the cloud. This paradigm, known as “hybrid cloud,” enables enterprises to intelligently burst their workloads to the cloud without sacrificing the use of their own hardware. Market Research Future προβλέπει that the hybrid cloud market will grow to $173.33 billion by 2025 as private cloud users shift to the hybrid model to avoid high costs and lower their security risks. But υβριδικά σύννεφα typically require having to create and manage data copies and app changes, which can negate — or even reverse — the cost savings.

Alluxio claims to have developed a solution in a platform that enables companies to orchestrate and serve data to apps in various environments including on-premises, cloud, hybrid cloud, and multi-cloud. Initially conceived as a research project at the University of California, Berkeley’s AMPLab, Alluxio sits between computation and storage in the big data analytics stack, providing a data abstraction layer for computation frameworks that enables apps to connect to various storage systems through a common interface.

“I was a Ph.D. student at AMPLab when I built the beginnings of Alluxio, originally called Tachyon, as part of my thesis. The goal was to build software to power all data-driven workloads, which includes large-scale analytics and AI, to significantly accelerate the data revolution,” CEO Haoyuan Li told VentureBeat via email. “Today, Alluxio is the data orchestration layer for critical data driven workloads, such as analytics and AI and machine learning, in the cloud.”

Connecting data sources

Hybrid clouds always include a private cloud — i.e., servers, storage, and networking resources behind a firewall and run on-premises — and are typically managed as one entity. A hybrid cloud app might draw from an on-premises database while running code both in the on-premises private cloud and scaling to the public cloud when demand increases.

These sorts of apps use APIs provided by Alluxio to interact with data from storage systems at high speeds. As Li explains, the platform is commonly tapped for AI and machine learning use cases to drive insights from large collections of small datasets.

Above: The Alluxio technology stack.

Image Credit: Alluxio

“A typical scenario is training based on many small images to drive recommendation engines for social media and gaming platforms. Commonly, a single enterprise is training numerous such models every day, each on billions of small files, using a compute infrastructure spanning a few hundred to a thousand servers with CPUs and GPU,” Li said. “Alluxio’s data orchestration platform is a ‘zero-copy’ bursting solution so when compute capacity is limited, users can intelligently burst to the cloud without needing to make costly data copies or app changes.”

Cost savings is a key piece of the puzzle. In a recent επισκόπηση from Pepperdata, a third of businesses reported going up to 40% over their initial cloud budget thanks to unexpected costs. A separate αναφέρουν by Anodot showed that businesses with more than $2 million in cloud costs were generally surprised by how much they spent, with the majority taking at least several days to detect anomalous surges.

“As the pandemic has forced more companies to move to a more comprehensive digital and online presence, it has … accelerated the pace of adoption of technologies that enable better exploitation of data, deeper digitization, [including the cloud],” Li continued. “The result is that we have seen wider and faster adoption of technologies like Alluxio’s that enable exploitation of corporate data for the benefit of the enterprise.”

To this end, the open source project on top of which Alluxio offers services, Alluxio Open Source, now has over 1,000 contributors. On the customer side, Alluxio says it has more than 1,000 users across eight of the ten largest tech companies, including Facebook (now Meta), Uber, Alibaba, Tencent, Bytedance, Baidu, and Amazon as well as Comcast, Rakuten, and China Unicom.

“As a data platform for both analytics and AI spanning multiple data centers and clouds, there is no technical solution which competes directly with Alluxio,” Li asserted. “Typically, our customers adopt us as a point solution first, and then grow into an enterprise wide data platform. As a point solution, there are other solutions competing with us. As a platform, we are in a unique leading position, and there are no real alternative solutions.”

Alluxio has a team of around 50 employees, which it plans to grow to over 100 by the end of the year. The startup has raised a total of $70 million in venture capital to date, mostly in equity.