Real-time data infrastructure platform “Timeplus” received seed round financing led by Hillhouse Ventures

Visit the original URL

36氪 was informed that the real-time data analysis infrastructure platform “Timeplus” centered on streaming data has recently completed a seed round of financing. This round of financing was led by Hillhouse Ventures. Other co-investors and expert advisors include Jeremy Kranz, investment director at GIC TMT; former Apple senior vice president Rory Sexton, who has been responsible for building Apple’s global supply chain capabilities for the past 20 years; Richard Tibbetts, real-time streaming database expert, now Tableau product Vice President; Margaret Lee, Senior Vice President, General Manager, Digital Services and Operations Management, BMC, and former Vice President of Splunk.

Timeplus, established at the end of November 2021, is a new generation of streaming data-oriented data infrastructure platform, focusing on real-time streaming data analysis.

After a little science, the current big data analysis systems can be roughly divided into two categories: batch big data and streaming big data. Among them, batch big data is also called historical big data, and streaming big data is also called real-time big data. There are differences in the processing time of different types of data. According to some information, the current complex batch data processing (batch data processing), the usual time span is between tens of minutes to several hours; interactive query based on historical data, the usual time span is tens of seconds to minutes; streaming data processing based on real-time data, typically spans from hundreds of milliseconds to seconds.

Wang Ting, founder of Timeplus, introduced to 36Kr that from the perspective of industry trends, the current real-time data represented by IoT/machine behavior, application behavior, and interactive behavior is growing the most rapidly, and is expected to exceed 30% in 2025. Gartner also predicts that more than 50% of enterprise-level applications require real-time data analysis capabilities to improve enterprise operational efficiency. At the same time, the overall actual utilization rate of data is less than 5%, the utilization rate of real-time data is lower, and the potential and scenarios that can be tapped are even greater. IDC also predicts that “enterprises are seeking a new generation of data infrastructure that can fully meet and embrace this real-time data.” From the perspective of the actual business role, a big data infrastructure that can meet the needs of real-time data utilization can improve the operational efficiency of enterprises internally, understand customers faster and better externally, and serve customers, so as to produce high-quality products that meet customer needs. .

Wang Ting introduced as an example that streaming big data infrastructure can help companies make better use of streaming in scenarios such as real-time supply chain visualization and logistics monitoring, smart manufacturing, real-time financial fraud, real-time marketing and personalized recommendations, and real-time monitoring of smart cars. effect of the data. For example, in the real-time monitoring scenario of a smart car, the real-time data generated by a smart car in a day may exceed 10TB.

Therefore, the real-time analysis and monitoring of these huge real-time streaming data, user driving and interaction behaviors can further ensure the driving safety of the car and improve the user experience. To further explain the logic, in the battery monitoring scenario of smart electric vehicles, on the one hand, it is necessary to quickly base statistics on some indicators in the real-time battery sensor flow data—for example, the number of times the voltage is lower than 11V in the last hour is used as the initial abnormal signal. To trigger real-time warnings, it is also necessary to compare with historical data (such as abnormal battery data such as weather, time period and road conditions) to further clarify or exclude some false alarms and details generated by the analysis of data that is too short.

According to Wang Ting’s observation, traditional data processing product solutions are difficult to effectively support these requirements in terms of performance, latency and flexible deployment. This is just an example, the same demand occurs in finance, supply chain, marketing, etc. scenarios. To sum up in one sentence, with the explosive development of data volume, complex and real-time business requirements such as personalized service, user experience improvement, intelligent analysis, and in-process decision-making have put forward higher requirements for big data processing technology. It is also a necessity for the further development of streaming data infrastructure.

Specifically comparing the products that have appeared, Wang Ting believes that most of the current data warehouses and analysis platforms are oriented towards full historical data, centralized and unified. Due to the large-scale and full processing of data, the overall processing time for demand is too long, and the freshness of data (data freshness) often cannot meet the user’s timeliness needs. Therefore, time-based real-time analysis of data and responses to real-time intelligence and automation will become a new industry trend.

In response to these pain points, Timeplus wants to build a new generation of time series-oriented real-time analysis platform for streaming data—that is, products that can aggregate real-time streaming data and historical data with high efficiency and low latency, and provide advanced analysis scenarios. . In terms of business value to enterprise customers, it hopes to help develop new real-time data analysis and real-time operation automation scenarios, reflecting the value of real-time data-driven business.

In terms of overall effect, a core indicator for judging the performance of a real-time streaming data analysis platform is the processing and analysis delay and throughput of real-time event data.

In the current test environment, the end-to-end latency of Timeplus’ real-time complex analysis is less than 100 milliseconds or even less than 10 milliseconds on ordinary machines, and the performance exceeds the industry level by more than 20 times. At the same time, the throughput of real-time event analysis can exceed 1 per second. Thousands of events. In addition, the overall deployment cost of Timeplus is small, and it can be flexibly deployed from the cloud to the end, even to the device side. “While completing powerful and complex analysis functions and excellent performance, the consumption of data computing and storage is far lower than the current industry solutions. .” Wang Ting said.

The reason why it can achieve this effect is closely related to Timeplus’s real-time data analysis architecture design based on stream-first and integrated vectorized analysis. Dismantling the underlying technical architecture, Timeplus has designed a unified real-time analysis engine with time as the core, supporting multi-layer computing models, and taking into account both streaming and historical analysis. The engine is designed to make the process of collecting data in real time, processing it in real time, and generating real time insights simple and fast.

First of all, in terms of speed, Wang Ting told 36Kr that Timeplus’ engine can quickly store real-time data and feed back its overall analysis speed. For example, in the process of importing real-time data into its streaming analysis platform, its engine can increase the import speed by 40 times compared to its peers. In addition, this engine can process real-time data and historical data at the same time, making the joint query of the latest data and historical data possible and efficient, and does not require a lot of pre-calculation and repeated calculation. It can be seen that the role of this engine is to import efficiently, and in addition, it also makes the overall analysis speed higher in processing. The engine is also one of the company’s core barriers, supporting the acceleration of overall data analysis at the bottom.

Moreover, Timeplus can also store data in a specific format, which makes data storage consistent and efficient. In terms of more specific effects, Wang Ting introduced: “The time-optimized new-generation data storage and analysis format TDF (Timeplus Data Format) will allow data to be stored in memory, disk or cloud, whether it is streaming analysis or Historical analysis and query can be highly consistent and efficient. Not only does the data store only one copy, but also achieves ultra-high storage and retrieval capabilities, high availability, no data loss, and supports vectorized analysis.” Reduce server storage costs and enable sub-second real-time analytics.

At the same time, in terms of ease of use, Timeplus chooses SQL as the unified analysis query language, which will allow most enterprise customers to quickly access data for analysis and exploration without having to learn and use new languages ​​or coding, and obtain real-time insights. Of course, using a standard language such as SQL will also make Timeplus faster and more convenient to aggregate data from different databases.

In addition, Wang Ting also emphasized that Timeplus is not only characterized by speed. It believes that the greatest value of Timeplus as a real-time data infrastructure platform lies in truly transforming data into real-time decision-making and real-time automation of enterprises, promoting huge business growth and enhancing enterprise competitiveness. In specific scenarios, Timeplus can help users to travel in time through its own products – that is, users can freely inquire about the statistical analysis situation a few hours ago. That is to say, such products are not free of technology and scenes. The big data processing and analysis platform for streaming data must first be able to meet the business needs of customers, promote the development of business in corresponding scenarios, and on this basis, improve product performance as much as possible.

v2_4d68c33cf8164c69afbfd85b1bf2134d_img_

Product Architecture

According to reports, Timeplus has officially released the beta version of the product, attracting the participation of some global customers and partners, such as AlphaStream, Aurora, datapm, gravitydata, etc. Take AlphaStream as an example, as a Brazilian fintech company, they used Timeplus to achieve real-time stock market pricing. The executive director of Alpha Stream said: “We were able to simply plug the source into Timeplus and start writing hot queries on the streaming data to get the results. No code to compile and deploy. We could also direct the results to the sink for use in the dashboard, or even Can be combined into another analysis. This enables prototyping to deploy applications very quickly.”

In addition to the products, Wang Ting also introduced the overall characteristics of the company. Specifically, the Timeplus team has international genes and accumulated deep industry production and technical experience. Its founding team comes from industry-leading data platform companies such as Splunk, SAP and Amazon. The duration of cooperation between teams reaches 10 years.

Among them, CEO Wang Ting is the former vice president of global engineering at Splunk. As the founder in 2012, he prepared and developed Splunk’s first overseas R&D center, leading a team of hundreds of people. Earlier, he served as the global vice president of product engineering at SAP BusinessObjects, responsible for the product and R&D of Crystal Reports and Dashboards, the main product lines of the world’s leading BI platform. part. In addition, Wang Ting also served as CTO at TalkingData. The company’s COO, William Plummer, graduated from Harvard Business School and has more than 20 years of experience in the world’s top strategic consulting companies such as Goldman Sachs. He was the chief strategy officer of TalkingData, responsible for the company’s strategic planning, management of overseas business lines, and KA customer and channel cooperation . Tao Gang, CTO of the company, has more than 20 years of experience in enterprise software design, and served as Distinguished Engineer and Architect of Huawei Canada. He is also the former chief engineer of the AI/ML and data ecological platform department of Splunk headquarters and the former chief architect of SAP business intelligence visualization.

Because of these backgrounds and experiences, Wang Ting said that the company will focus on global markets such as Europe and the United States in the next period of time. Also due to the accumulation of the company’s team overseas for many years, it already has customer leads in the financial, supply chain, industrial Internet and Internet industries. In addition, the company also has a R&D team in China, is recruiting new members to join Timeplus, and explores business opportunities with partners when the time is right.

In the near future, Timeplus will also release official products in order to reach and serve more customers.

About investing:

Li Qiang, a partner at Hillhouse Ventures, said: “Global data growth has entered a stage of continuous explosion. The total growth in the next three years will be the sum of the past 30 years, and streaming data accounts for the highest proportion – reaching more than 30% of the growth. , and the existing infrastructure is difficult to cope with the demand under the rapid growth of this data. The founding team of Timeplus has more than 15 years of practical background in the software industry. Based on years of engineering practice and technological innovation, the cloud-native streaming data analysis platform created can help enterprises achieve higher business agility. We are optimistic that Timeplus has the opportunity to become a next-generation data analysis infrastructure, which is why we support Timeplus from the seed round. It is believed that under the leadership of founder Wang Ting, Timeplus will provide continuous innovative data service capabilities for more industries and enterprises with top technology.

media coverage

36Kr Investment Community Investment Network
Related events

This article is reprinted from: https://readhub.cn/topic/8htuLY9MP0J
This site is for inclusion only, and the copyright belongs to the original author.

Leave a Comment