Pliops lands $100M for chips that accelerate analytics in data centers

Operations may be made better by analysing data created inside the company, such as sales and purchase data. However, some firms are having trouble effectively processing, storing, and using their massive volumes of data. Seagate commissioned an IDC study that found that just 57% of the data accessible across all of their business lines is actually collected by businesses.

Data-intensive applications need a lot of resources, and constructing the requisite computing and storage infrastructure is often costly. According to IDG, businesses who are particularly shifting to the cloud this year want to invest $78 million in infrastructure. 36 percent of respondents named cost management as their greatest issue.

Uri Beitler founded Pliops to provide what he refers to as “data processors” for business and cloud data centres. According to Pliop, his processors are designed to increase the speed of databases and other programmes that use flash memory, which will ultimately result in cost savings.

“It quickly became apparent that the design of today’s data centres is incompatible with the data demands of yesterday. Massive data expansion has ran afoul of dated compute and storage flaws, resulting in processing sluggishness, storage bottlenecks, and declining networking effectiveness, Beitler told TechCrunch in an email interview. Despite the fact that CPU performance is rising, it is not keeping up, particularly in situations when accelerated speed is crucial. Increasing infrastructure is often both expensive and challenging to maintain. Organizations are searching for methods to liberate CPUs from computationally demanding storage chores as a consequence.

Pliops is not the first company to release a data analytics processor on the market. The BlueField-3 data processing device is offered by Nvidia (DPU). There is Octeon technology from Marvell. The SPARC M7 CPU from Oracle contains a data analytics accelerator coprocessor with a unique set of data transformation instructions. Startups like Blueshift Memory and Speedata are also developing technology that, according to their claims, can do analytics jobs far more quickly than conventional CPUs.

Pliops lands $100M for chips that accelerate analytics in data centers-1

Pliops, however, claims to be farther ahead than others, with installations and pilots with clients, including fintechs, “medium-sized” communication service providers, data centre operators, and government laboratories, while without naming them. Investors seem to have been charmed over by the startup’s early traction since they invested $100 million in its Series D round, which ended today.

The tranche, which was headed by Koch Disruptive Technologies and included SK Hynix and Lip-Bu Tan of Walden International, brought Pliops’ total cash raised to date to more than $200 million. According to Beitler, the money will be used to develop the company’s hardware and software roadmap, strengthen Pliops’ partner network, and hire more people from outside.

“Many of our clients had remarkable growth during the COVID-19 epidemic, in part because of their capacity to respond rapidly to the altered working environment and the unpredictable circumstances. Pliops undoubtedly did. We were not impacted by supply chain problems, but some of our clients were,” said Beitler. We do not perceive a stop in the expansion of data or in the need to use it. Prior to the most recent fundraising round, Pliops was powerful; it is now much stronger.

Accelerating data processing

Pliops was formed in 2017 by Moshe Twitto, Aryeh Mergi, and Beitler, the former director of advanced memory solutions at Samsung’s Israel Research Center. Before joining Pliops, Mergi co-founded a number of firms, including two that were bought by EMC and SanDisk. Twitto was a research scientist at Samsung researching signal processing technology for flash memory.

The Pliop processor offers SSD drive fail protection in addition to in-line compression, a method that reduces data size by identifying similar data sequences and preserving just the first sequence. Beitler asserts that his company’s technology can increase capacity while decreasing drive space by mapping “variable-sized” compressed things into storage to eliminate unused space.

The hardware-accelerated key-value storage engine is a crucial part of the Pliops processor. Key-value engines handle all permanent data directly in key-value databases, which are databases where data is stored in a “key-value” format and enhanced for reading and writing. Beitler argues that since these engines often operate at high CPU utilization levels, programmes do not fully leverage the advantages of SSDs.

“Organisations are searching for methods to liberate CPUs from computationally expensive storage operations. By using a new generation of hardware-accelerated data processing and storage management technologies, which offers orders of magnitude improvements in speed, reliability, and scalability, our hardware contributes to the development of a contemporary data centre architecture, said Beitler. In summary, Pliops makes it possible to maximise the return on current infrastructure expenditures.

Also Read: Best URL Shorteners: 8 Best URL Shorteners to Shrink and Track Links

The processor from Pliops got marketed last July. According to Beitler, the development team’s current priority is speeding up the data import for machine learning use cases, which have become more popular with Pliops’ present and prospective clients.

Pliops lands $100M for chips that accelerate analytics in data centers-2

The road ahead

Pliops undoubtedly has its job cut out for it. After investing years in the development of its BlueField range, Nvidia is a strong contender in the market for data processing accelerators. Additionally, AMD signalled its broader intentions by purchasing DPU supplier Pensando for $1.9 billion.

Joining the Open Programmable Infrastructure Project (OPI), a relatively new initiative under the Linux Foundation that seeks to establish standards for data accelerator technology, is a decision that might pay off for Pliops. Pliops isn’t a member yet, but it makes sense that joining would expose its technology to a wider client base. Current members include Intel, Nvidia, Marvell, F5, Red Hat, Dell, and Keysight Technologies.

When questioned about OPI, Beitler declined, but said that the market for data acceleration is still developing and expanding.

Beitler said that the infrastructure and application teams “continue to see both teams being burdened with underperforming storage and swamped apps that aren’t fulfilling the company’s data expectations.” The consensus is that our processor is a game-changing device, and without it, businesses would have to devote years of resources in hardware and software engineering to address the same issue.

Leave a Reply

Your email address will not be published. Required fields are marked *