• CCTC 2017 China Cloud Computing Technology Conference

    May 2017 18-19, the domestic cloud computing technology field of the most professional and most influential event - China Cloud Computing Technology Conference CCTC 2017 in Beijing, Chaoyangmen, Tang Tang Crowne Plaza Hotel grand convening. More than 60 lecturers from home and abroad, as well as more than 2000 cloud computing professional developers and related practitioners participated in the event.
    Nowadays, enterprise "cloud" pace is accelerating, especially in the new wave of artificial intelligence technology as the representative of the promotion, on the one hand enterprises have enhanced their data storage, computing and intelligent connection application ability through cloud technology; on the other hand, the use of new technology, big data, cloud computing on artificial intelligence based on the enterprise, but also can efficiently find out the data and method of improving business at low cost, and implementation of cloud, number and wisdom of natural fusion and development cooperation. [detailed]

Click view: Live video

Graphic memoir

  • May 18th
  • May 19th
  • 17:30[container and operation and maintenance technology summit], the United States group review container platform person in charge of Zheng Kun issued the "United States group review business container actual combat sharing" keynote speech. Zheng Kun first introduced the United States Group cloud introduction container reasons, due to the business cycle and sudden, core business to x times, capacity redundancy. The traditional virtual machine deployment service has three low efficiency problems: 1. manual expansion, high degree of artificial participation; 2. complex process, poor reliability; 3. business reserve machine, waste of resources. So the cloud platform container design strategy for the upgrade of existing infrastructure based on the evolution of; grasp the rhythm, the "stability" first; to do architecture, and performance optimization. Zheng Kun detailed the design details of the container platform, developed a container support service governance solution, and finally shared the problems encountered in containerization practices and solutions.

  • 17:00[block chain summit] [how to learn block chain technology] round table start, interactive link still by Shanghai music live CTO Wang Wei presided over, he put forward: "now the block chain technology background is what?" Domestic technicians learn the school district block chain? How is the school district block chain? People's insurance, Li He, silver chain technology, CEO Shen Tu youth, Haina cloud CTO Zou Jun, "integral coins" founder Zhang Lei on these technical people are most concerned about the problem, carried out a wonderful answer.

  • 17:00[big data core technology and application summit] iFLYTEK real-time computing platform senior architect Liu Jun in "Ifly speech cloud real-time analysis framework and the practice of" the theme of a speech that Ifly speech cloud can provide speech recognition, speech synthesis, face recognition, voice recognition, semantic understanding, full duplex with context AIUI voice interaction ability. With the development of business, we put forward a real-time demand more, Ifly speech cloud can be real-time statistics of key indicators, real-time monitoring and other functions of the cluster, the health service. Then, Liu Jun shared the three major practices of voice cloud in the field of real-time analysis: BI& large screen, real-time monitoring platform, cloud know. The real-time monitoring platform supports various alarm categories, such as server basic index, monitoring alarm, active alarm and so on. The alarm trigger condition, alarm time and alarm notify personnel are configurable.

  • 16:50[Spark Technology Summit] Databricks Software Engineer Fan Wenchen shared the title entitled "Deep Dive: How Spark Uses Memory" speech. Fan Wenchen's speech mainly includes three aspects: how to look at memory management under Spark cluster, how to deal with the problem of memory competition, and how to optimize the memory usage of Spark. He explained the problem of memory competition often seen in practice, when Storage and Execution competition priority Execution, consideration of priority to the Storage memory to brush off. In addition, there are between Tasks distribution, not static advance allocated, but through the dynamic run-time processing competition, and by way of cooperation and let the memory user cooperation, so we try to solve the memory algorithm by fair competition problems.

  • 15:40[the core technology of big data and the actual application of the summit] HanSight Han Si senior technical engineer Zhang Jun in the "big data security analysis" in the exploration and practice of electronic banking in a keynote speech said with more safety data, information security problems will be transformed into a large data analysis problems, safety of large-scale data need to be effectively connected analysis and mining. Specific to the electronic banking industry, in the face of a large number of daily visits and transaction logs, we need to build a user profile with large data analysis, including library collision behavior, portrait and scanning behavior portrait. HanSight big data in electronic banking security analysis has a deep exploration and practice, and analyzes many details of the platform architecture design, such as real-time distributed stream processing platform for large-scale distributed real-time processing system based on Flink development; the built-in data dictionary all the input data for Elasticsearch regularization etc..

  • 16:30[container and maintenance technology summit] hyper.sh operation for Pei Tong lecture on the theme "Hyper container cloud and cloud operation and maintenance", Docker is the most popular container technology, which is mainly composed of Container (lightweight, fast) and Docker (Image, portable, whenever and wherever possible no change). However, there are security concerns, normal use should be safe, better in the virtual machine. You can make the virtual machine faster by speeding up startup and reducing memory footprint". The advantage of Hyper Container for public cloud is that the scheduling efficiency is higher and the management complexity is reduced. The "cloud" in the Docker experience and on your computer using the Docker agreement, but the actual operation is the cloud resources -- as users have a limitless resources, according to the "host" just need to create / use / container can be destroyed, but not for the bottom of the "host" or "clusters". Finally, Pei Tong shared his own thoughts from 4 perspectives: resource perspective, environment / configuration management, application change, and Metrics information collection / monitoring.

  • 16:30[block chain summit] integral coin founder Zhang Lei, shared the "block chain business analysis method practice" keynote speech. Zhang Lei first elaborated the concept of block chain business analysis in public, including Blockchain, Business, Analysis, and IIBA based BABOK 3 standards. Zhang Lei thinks: "what use database system to be able to offer mature solution, use block chain to be very foolish.". Block chain represents a new way of trading, in part, by giving up part of the security rights in exchange for centralized rights.

  • 16:10[Spark Technology Summit], Alibaba technology expert Peter Chen shared the speech entitled "SparkSQL's application in ETL". Song Jun said, ETL has three main steps: extraction, conversion, loading. First read the data source, do the washing processing, integrated processing, finally put these data to the target storage, to meet the requirements of ETL is simple and easy to use, support a variety of data sources, support fault-tolerant processing, rich operator, complex data types, fast calculation etc.. How to achieve these goals, from DataSource, rich operators, Hive compatibility, performance, cloud ETL five aspects of interpretation.

  • 16:00[block chain summit] Shenzhen silver chain technology CEO, gold chain League Deputy Secretary General Shen Tu youth issued the "how to build block chain application" keynote speech. The current application areas of block chain include cross-border payment, electronic bills, commercial bank collateral and so on. In the application development of block chaining, the application scenarios should be considered first, then the selection of public or private chains, and then the selection of development tools. The system architecture of block chain development includes user, core enterprise block, chain management platform, block chain middleware and Fabric block chain network. The specific development process is to build Fabric nodes, to write intelligent contracts, to achieve intelligent contract interaction, and to implement JAVA SDK.

  • 15:40[big data core technology and application summit] gold electric LaSalle chief technology officer shangf said in his keynote address, after analysis, big data can have many typical applications, but in reality because of the bottleneck of knowledge, talent, technology, data, data applications landing facing a lot of problems. As for how to deal with and make breakthroughs? Shang Feng proposed that data processing methods should be promoted from data manual workshops to intelligent data factories. Intelligent data factory, using the "data" and "experts" double drive mode, relying on industrial production, human knowledge, breakthrough the bottleneck, fast, high efficiency and high quality to adapt to the new data, adapt to the new business. As to how to build an intelligent data factory? He has carried on the detailed analysis from the frame design, the business process, the technical flow and so on many angles.

  • 15:30[container and maintenance technology summit] where a Thai geeks research vice president Yang Tao share the theme of "financial institutions for vertical container delivery network", CALMS is not GPS, but CALMS and compass DevOps workflow others, just provide a guide rather than an exact answer. CALMS's focus on DevOps is culture, not perfect. It's just a guide, not the final answer. Build a small and agile team that has reached a strong agreement (for example, some requirements documentation and API). This allows the team to actively correct code segments as long as they do not violate agreements that tie teams and code together. Yang Tao then explained in detail the social Coding (Social), tool chain, delivery network -ChatOps.

  • 15:00[block chain technology summit] Haina cloud CTO Zou Jun published a brief history of consensus speech. Distributed computing is a computing model in which each node communicate and coordinate actions by sending messages to each other in order to achieve a common goal. Because the nodes of the distributed system run independently and communicate with each other, they will fail. Therefore, to ensure the consistency of the whole system, consensus protocols are needed, and the distributed systems guarantee consistency through these two points. At present, most of the consensus of block chain is a single scene, limited by FLP, CAP. Finally, he concludes: "block chain is trust machine, consensus algorithm has strong consistency and final consistency consensus algorithm. Chain block consensus algorithm is different from chain, union chain and private chain.

  • 15:00[Spark Technology Summit] pleasant loan data scientist Wang Ting shared the theme of "Spark practice in the context of financial fraud" speech. Wang Ting first introduced financial technology companies face fraud risks, including user repayment ability, false application and fraudulent transactions, etc., the traditional approach to solve such problems is often inefficient. She thinks anti fraud is also a machine learning process. Based on massive data, Yixin constructs a financial knowledge map - FinGraph, which has three aspects: application scenarios, system support and data integration. Wang Ting focused on the anti fraud scene in Spark Spark Streaming SDK with three axes: real-time data processing; Spark GraphX for mining and mining engineering drawing features Spark Mllib fraud group; used for Feature Engineering, simple machine learning training. Wang Ting focused on the architecture of Spark platform based on real time anti fraud and pleasant credit development, including user layer, tool layer, model layer, data processing layer, storage layer and data layer, according to the different events that real time fraud score. In addition, the pleasant loan through the introduction of anti fraud investigators, combined with Atlas mining, improve feedback efficiency, rapid detection of fraud gang.

  • 15:00[container and maintenance technology summit] Shenzhen Rui Yun Chi CTO Xu Niangang share the theme of "new normal" container of the leading IT speech, introduced the new normal development of IT software products, ten years of delivery change, organization structure transformation and the cooperation way, Bimodal is becoming the traditional enterprises to adapt to the new normal and IT the model of governance, the application container will be the next most popular software movement "". Many of the technologies related to containerization are not new technologies, nor are they the technology of racking their brains, testing, containers providing infrastructure for applications, research and development, and delivery to the complete chain of operations and maintenance. The advent of container technology is also contributing to the development and landing of other technologies. Container technology is not only the sublimation of traditional IT technology and ideas, but also new ideas and ideas from Internet IT practice. The container practice and import of traditional enterprise IT is a process from the old IT architecture to the new IT architecture. Xu Niangang said the container technology is still a "moving fast" in the stage, the container is a process from the point and surface process from reliable to innovation, the traditional enterprise can try to start from the package, the deployment process to the last comprehensive container platform hosting applications such a gradual introduction. Container technology has challenges from application, technology and ecology.

  • 14:20[block chain summit] Li Yansheng, chief architect of IBM labs Services Division in Greater China, delivered a keynote speech on Hyperledger technology practice and case practice. The blockchain key sharing books, including smart contracts, privacy protection, consensus, and the blockchain value is to save time, reduce costs, reduce risk, supervision organization, business organization and operation mechanism. Li Yansheng analyzed two business cases of business process and business process on the block chain. Finally, he summed up the technical challenges of block chain applications: business challenges promotion mode 1. commercial network; 2. commercial network organizational structure and management methods should be formulated; 3. supporting social rules; 4. increase the technology stack brings new technical challenges; the consistency problem due to the data of the 5. distributed system design; 6. data logic and business logic; 7. system deployment complexity issues more prominent; 8. of the infrastructure platform and standardization requirements are higher.  

  • 14:40[the core technology of big data and the actual application of the summit] travel drops a senior software development engineer Li Yang in "HBase in the application scenario and best practice" by travel theme speech that HBase online business, offline business has been applied to travel in drops. The main data types stored in the HBase include statistical results, report data, raw fact data, intermediate data and result data, and backup data on the online system. In view of these four data, Li Yang cites an application scenario, such as order event, driver passenger track, ETA (estimated arrival time) model, real-time training and monitoring tools. Each application scenario resolves the application requirements, implementation process, and the like.

  • 14:00[big data core technology and application summit] Beijing University computer department associate professor Xiong Yongping in the "INFINIDATA" in his keynote speech said: platform enterprise data application of unified data management and data processing based on Spark, including the common needs and complex SQL query and retrieval, a full table scan, interactive exploration. MPP data warehouse and big data platform are two major solutions now, but it is difficult to meet the needs of enterprises because of the high price and difficult to use. From the user's needs, INFINIDATA platform came into being. INFINIDATA has integrated data management, easy to use and powerful computing flow management, and provides standard and high data inquiry service. In the field of visualization, INFINIDATA has also been explored, mainly focusing on row, column, value correlation analysis, discrete matrix analysis, OLAP analysis, visual analysis. At present, INFINIDATA has many typical applications in the insurance industry, such as list inquiry, statistical reports, customer cleaning and portrait, etc..

  • 14:00[container and operations technology summit], Mesosphere co-founder and CTO Tobias Gunter Knaup, and Asia Pacific Technology Director Ben Lin to share the theme of "DC/OS 1.9, DEEP, DIVE.". Tobias represents the framework of enterprise partners that can easily run and deploy data services operations over DC/OS, simplifying the management of DC/OS clusters. Mesos has been looking for and allowing more workloads to run above DC/OS and to implement new use cases by adding POD and GPU. DC/OS supports many canonical critical applications. Mesosphere DC/OS provides powerful capabilities like AWS platform services and can be run on any infrastructure.

  • 13:45[Spark Technology Summit] seven cattle cloud senior Big Data Engineer Zhao Hongyao share entitled "Spark based streaming processing engine in Pandora big data products" speech. It is reported that Pandora's current data processing scale has reached the amount of data written in real-time every minute to hundreds of GB, real-time data written every minute billions of entries. Zhao Hongyao introduced the Pandora core components - computing platform architecture, including real-time tasks and offline task scheduling management platform, resource management platform (YARN, Mesos, self research container cloud platform) and Spark components. He focuses on two issues that need to be taken into consideration: the user's use of the interface, and the shielding of technical details. User interface, he suggested that schema has the characteristics of robust data flow, efficient storage and computing, and rich data types. In the technical details of the deal, the common problems in the production of Spark Streaming operation management and monitoring, data loss, data consumption, delay aggregation operation processing, Zhao Hongyao gives the solutions.

  • 13:40[block chain summit], founder of the Po, talk about block chain tutorial, author Tan Guopeng published a thorough understanding of cross chain technology keynote speech. The purpose of cross chain is to realize the asset transfer between different chains; swap (atomic); to solve the problem of Oracle; to realize the pledge of assets and the reading and verification of its chain information or events. At present, cross link technology mainly consists of three technologies: one is the fairness of single or more entities, the other is side chain technology / relay technology, and the other is hash lock. Justice is a simple technique, but the downside is that you need to trust 1 or more entities. Finally, he analyzed common cross chain projects such as Interledger and polkadot.

  • 13:30[big data core technology and application in the Apache HAWQ summit] founder Lei often under the auspices of the official start.

  • 11:40[block chain summit] [round table dialogue: enterprise opportunities on block chain innovation] round table dialogue, moderator Wang Wei and three guests for block chain applicable industries were discussed. Among them, Li he said that in the application context of the intermediary is the most appropriate application, such as the insurance industry. Dong Ning said: "different industries on the use of block chain is not the same, we should consider in accordance with their own enterprise application scenarios, consider the value of some block chain.". Zhang Zhengwen believes that the development of the block chain for the financial industry, because the current national regulation of the financial industry mainly solve the problem of trust everyone, and attribute to the center of the block chain reduces the cost of supervision, solve the problem of trust.

  • 11:40[container and operation and maintenance technology summit] where to go to the network platform Division Development Engineer Xu Lei speech, the theme for "Mesos/Docker based data processing platform.". Where data processing platform to handle about 34 billion /25TB of data every day, of which 90% of the data processed in 100ms, the longest 3h/24h data playback, private Elasticsearch Cloud, automatic monitoring and alarm. The platform chooses Mesos because it is simple, stable, and has a more mature scheduling framework. Storm, Spark, Flink, and ELK are sent to the Mesos via Docker. Xu Lei introduced the container's main points of attention for JVM, and the configuration parameters and run parameters that Mesos containers need to pay attention to. Spark on Marathon realizes standardization of operation and maintenance & automation, and solves the deficiency of Mesos-Dispatcher. The problems are Checkpoint & Block, temporary file, Coarse-Grained. Elasticsearch demand gradually increased, currently hosting the 40+ cluster, 100TB+ business data, peak 1.2k, QPS, about 110 nodes. Subsequently, Xu Lei shared where to go in monitoring and operation and maintenance experience.

  • 11:35[big data core technology and application of Apache HAWQ Committer summit], PMC, Beijing science and Technology Co., even architect Tao Zhenglin published "the era of big data cloud management engine HAWQ++" keynote speech, Tao Zhenglin introduced the HAWQ and its development, then introduces the HAWQ architecture, HAWQ architecture, HAWQ optimizer, internal HAWQ query processing, HAWQ resource management, HAWQ storage, HAWQ++ new features, HAWQ++ installation deployment, how to access HAWQ++ and HAWQ++ import and export data and HAWQ++ combined with the use of MADlib and so on. HAWQ the goal of resource management is responsible for the resources and resources for HAWQ users and back to YARN, and query operators allocate resources; three level resource management including global resource management, internal resource management and operator level resource management; multi-level resource management and the management of CPU and memory. In the HAWQ++ import and export data, you can import export data, gpfdist external tables, import and export data, COPY commands, import and export data, and import data into the hawq load tool through the HDFS external table.

  • 11:20[Spark Technology Summit] HUAWEI big data design minister, Apache Carbondata PMC Committer, entitled "CarbonData big data high performance interactive analysis practice" speech. He first analyzes the enterprises face in the construction of data platform challenges, several schemes for both NoSQL, parallel search engine, SQL, or on Hadoop, are only for a scene design, can only solve part of the problem. The architect has to compromise on system design, or only partially apply; or copy multiple data to meet all applications. And CarbonData's goal: a data to meet a variety of business needs, and seamless integration of large data ecosystem. CarbonData entered the Apache incubator in June 2016 and had its first production system deployment in September 2016 and became the top Apache project by April of this year. The latest version now is 1.1.0, and the computing engines that access CarbonData are Spark, Presto, Hive, and Flink. Currently, HUAWEI, the U.S. group, SAIC, Hulu and other companies have deployed CarbonData in the production system. Li Kun explained in detail several important technical advantages of CarbonData. Query: two level index, reduced IO, suitable for ad-hoc query, any dimension query scene; delayed decoding, vector processing: for a full table scan and analyze the scene; data management: incremental storage, multi-level sorting can be adjusted by the user balance storage time and query performance, incremental updates, batch merge, fact table or dimension table supports fast update, leisure to do with Compaction; mass: computing and storage separation, support from GB to PB mass data, data trillion second response; the deployment of Hadoop Native format, and large data ecological seamless integration, using the existing Hadoop cluster assets.

  • 11:00[container and maintenance technology summit] China Mobile Su Yanyun computing products PaaS Product Manager Cao Gaojin shares entitled "China Mobile PaaS platform technology selection and experience sharing" speech. Independent planning and construction business system, there is no long-term independent R & D team IT lead developers are not unified architecture, application of chimney type architecture, universal application, Rock Island system, lack of public power of abstraction and reuse, system function caused by repeated construction, the Internet under the new norm, "seckill", grab a red envelope on the impact of architecture and application. China Mobile Su research for its own circumstances, built the elastic PaaS, 11 introduced some of the problems encountered in practice. Finally, Cao Gaojin has the following experience: PaaS enterprise need some preparation mechanism: Service Catalog, Application Catalog, Cloud Provider Interface; virtual machine and the container has the same requirements; OpenStack + Kubernetes may cover the enterprise really want.

  • 11:00[block chain summit] Chi chain technology CEO, former IBM Asia Pacific IT economics senior consultant Dong Ning issued the "enterprise class block chain development platform" keynote speech. He first narrated the enterprise to use the block chain the necessity as well as the block chain circulation value expression. The enterprise needs to consider the block chain platform technology extension and industry perspective, block chain as the technology sector segments, combined with each other will be in the future and other development technology sectors, including cloud computing and artificial intelligence, networking and so on, so the future will form a "block chain +, block chain technology platform through consumer supply chain, manufacturers of information chain, make the circulation of value is more credible, accelerated liquidation business market data. Finally, Dong Ning said: "the ability to use block chain technology platform to help businesses focus on business.".

  • 10:40[big data core technology and application summit] Hortonworks technology expert Liang Yanbo delivered a speech for the 1 billion level "large-scale machine learning" feature of Spark based on the theme, the share is divided into five aspects: first, Liang Yanbo introduced the application background of Spark scale based on machine learning; secondly, introduces the application of Spark vector based on L-BFGS; thirdly, share fourth carrier free L-BFGS logistic regression, based on a detailed analysis of its performance; fifth, introduced the integrated with existing MLlib; finally, the future research direction of share. Through a large number of examples, Liang Yanbo said that in the future there are still a lot of research direction, such as reduced iteration, improved performance and prediction of ad click rate and the formation of real cases with billions of parameters, we also share the experiences and lessons of income etc..

  • 10:20[block chain summit] distributed technology CTO& founder Zhang Zhengwen published the "block chain intelligent asset development practice" keynote speech. From electronic cash to intelligent contracts to intelligent assets, block chains continue to evolve, and bitcoin related categories are too large, giving birth to the birth of an Ethernet square, where smart contracts run on the etheric square, But the aether workshop is a little flashy. The small ant block chain is a block chain, digital asset and application platform. Digital assets come into being with the birth of the system. It is an electronic contract with digital identity. Digital asset life cycle including registration, issuance, transfer and destruction, this is a small ant running mechanism of intelligent contract, you can use the Java, C++, Python and other languages based on the development of intelligent small ant contract, also can develop small nodes based on ant.

  • 10:20[container and operations technology summit] Citrix senior technical manager of Greater China He Haoxiang delivered a keynote speech, "container platform cloudy platform operation and maintenance practice.". Containers can be cross platform, mixed deployment, multiple platforms have different load balancing, but this only meets the 4 layer load balancing, can not solve the problem of application. The new type of application has the characteristics of unstable transaction volume, no location, no user patience, third party access, security and so on. From the long-term solution, the load balancing needs to be able to solve the new state application and deployment requirements, and the software design should meet the requirements of high availability, security, management, automation and analysis. He Haoxiang finally gave a detailed account of Citrix's concrete practices and demonstrations.

  • 10:20[Spark Technology Summit] Sina micro-blog advanced algorithm engineer Wu Lei shared the title "Spark based large-scale machine learning applications in micro-blog" speech. He said that micro-blog's current major challenges in large-scale machine learning technology is that the data feature categories, large dimensions, model diversification, large parameter matrix, resulting in a large number of memory and network overhead. Therefore, micro-blog R & D team developed the parameters of Spark server based on the use of asynchronous stochastic gradient descent algorithm, the main principle is the data block, block data pull parameters, calculation block data will be updated to the gradient and gradient parameter server. During the process, the performance of Batch, Size, PS, server, count, Sparse, Partitioning, Spark, memory and tuning modules of the parameter server is continuously optimized. In the process of solving micro-blog machine learning problems, they developed a unified computing framework called weiflow. Wu Lei explained the implementation process of weiflow one by one, through this framework, micro-blog team business development efficiency, business code contribution, execution performance, model performance and other aspects have been greatly improved.

  • 09:45[big data core technology and application summit] Computer Department of Tsinghua University associate professor Chen Kang published "optimization technique exploration" diagram calculation of the keynote speech, its contents include four aspects, first of all, on the map calculation are introduced briefly; secondly, to explain the use of the Architecture Department of the Ministry of acceleration calculation; thirdly, explain the 3D division figure acceleration calculation; finally, the core figure calculation acceleration method of share. The data sources of graphs are very extensive, and both Google and Facebook contain large amount of picture resources. Therefore, the computing methods are changed from data parallel to graph parallel. Chen Kang said that the current figure computing presents four characteristics: high access computing ratio, partial data is not good, the structure is irregular and driven by data. As a result, optimizing the speed of data loading is of top priority. Subsequently, Chen Kang introduced the calculation framework of computing systems, including the calculation of frame, such as easy programming, performance extension and automatic fault tolerance; computing framework to vertex centered and edge centered computing framework etc.. In the use of architecture, local acceleration diagram calculation, three-dimensional division of the graph, accelerated computing and the acceleration of the calculation of the external graph, Chen Kang used a lot of examples, and brought detailed technical sharing for the site developers. Finally, Chen Kang summarizes related methods of architecture optimization calculation that the figure calculation due to its wide application and the expansion of the scale, it is still a hot research content; the structure related method can accelerate the calculation operation system, such as the use of local structure and accelerate the division of 3D graph computation, graph the accelerated method of calculation and storage calculation; different pattern calculation, different methods to accelerate.

  • 9:40[block chain summit] people insurance information technology engineer Li he published the "block chain principle and application of insurance thinking" keynote speech. The block chain infrastructure consists of data layer, network layer, consensus layer, incentive layer and application layer. In the application of the block chain should pay attention to several points: first, the block chain and intelligent contract implementation, existing IT system can be realized, the block chain is to intermediary; second, block chain is not to achieve performance improvement, but to change the business model, the opposite performance decline; third, only to realize the endogenic chain the information on the external information into trust, unable to establish trust; fourth, block chain applications don't need money. At present, the most important application areas of block chain are the metal parts of the financial industry and various industries. Block chain based mutual insurance has the characteristics of low cost, post payment, post pricing and deadline fragmentation.

  • 09:30[Spark] IT Technology Summit Su ningyun headquarters senior technical manager Chen Ze shared the theme of "Spark App" analysis and diagnosis of automatic speech. Chen Ze first talked about Suning large data platform is divided into offline calculation, flow calculation, OLAP engine of the three part, based on the Spark application in which more and more widely used, the cluster size is 700 node Yarn cluster, the scale of the task 5W task / day, the amount of data processing of 300TB / day. Chen Ze focuses on Spark based CBT scheduling platform, SSMP platform and machine learning platform. According to reports, Suning Spark platform has encountered many problems in the process. At the operational level, awareness of Spark SQL is not enough, resulting in excessive dependence on Spark RDD level API of memory; calculation of misunderstanding, unreasonable use of Cache mechanism, resulting in a waste of resources; in the platform, there is no data to guide users to optimize the number of Executor and memory parameters, which leads to the problem of platform resource utilization rate is too low etc.. To this end, Suning developed Spark automatic App analysis and fault diagnosis system of Hua Tuo, the operation experience and knowledge tools, and with the index and events of two kinds of information, analysis and fault diagnosis of the task. Hua Tuo's current analytical and diagnostic capabilities include three dimensions of resources, performance, and failure, detailing the latest advances in each of these capabilities.

  • 9:30[container and maintenance technology summit], Ali cloud senior expert Chen Menghui share the theme of "Swarm evolution and large-scale applications", container layout field by MESOS, DC/OS, Kubernetes, Swarm three world. Swarm is an important product following the Docker Engine after the Docker, Chen Menghui introduced the Swarm architecture, API and scheduling from three aspects, and summarizes its advantages: the deployment of simple, efficient and friendly user interface, flexible constraints and affinity description, and its deficiency. Swarm Mode integrates Swarm cluster management and container scheduling functions into Docker Engine, provides Service level abstraction, and comes with load balancing. Swarm Mode has the advantages of special deployment, high availability services, Secure, by, default and so on, but only Service level abstraction, Stack level abstraction still has no API, and deployment of stateful services is more complex. Finally, Chen Menghui described in detail the application of Swarm in Ali, Swarm unified management cluster, Ali cloud proprietary cloud output.

  • 09:10[Spark] technology summit host by seven cattle Cloud Technical Director Chen Chao Chen Chao, briefly reviews the history of Spark summit in four years, with more and more Spark landing project in China, this one period of enlistment received more than 100 copies of the Spark theme of the speech application, through the audit committee selected seven elite lecturer and classic cases from them, including Su Ningyun, micro-blog, HUAWEI, Sina's seven cattle cloud, Alibaba, pleasant loans, Databricks lecturer will bring wonderful share in the field.

  • 09:05[big data core technology and application summit] Wang Hu dongfangguoxin large data center director issued a "big data core technology in the application of speech and practice" operators of the theme, the share mainly includes three aspects, first, the big data architecture operators; second, the core technology of big data system; third application and practice of operators. Wang Hu introduced to the past operators of large data platform Iaas/Paas/Saas is divided into three layers, and the future operators of large data platform will go deeper evolution, including the application domain, data domain, technical domain, domain based, security domain and open domain. After years of construction of operators, the current technology is mainly based on "open source + independent" research and development combined with the use of big data core technology to build business oriented applications and platform applications. The Hadoop version of Orient Guoxin is enhanced based on the open source version and is compatible with the open source version and can be upgraded as the open source version is upgraded. Based on open-source community source code, to achieve the two development, converted to its own core technology, and gradually the application and production practice validation part of the code submitted to the community, such as: K8s+Docker source code revision. In big data core technology in the application of operators business practice, operators of enterprise operation and management, around the big data as the core, for customers and internal staff, to achieve enterprise business operations and management. In the practice of business application, Wang Hu introduced five core technologies: data, decision, marketing, customer and performance. Finally, the tiger on the core technology of big data in the future development of operators is discussed, mainly includes six aspects, such as online real-time ability, ability, learning ability, opening ability, ability and application ability of the data.

  • 9:00[container and operation and maintenance technology summit], the morning of the Hyper Cofounder & CTO Wang Xu under the auspices of the official start.

  • 9:00[block chain summit] the block chain technology summit was officially launched under the auspices of CTO, a resident of Shanghai. He led the audience to review the historical development of block chain technology. In 2008 published "Nakamoto bitcoin: electronic cash system" a point to point, put forward the data structure of BlockChain, the official opening of the bitcoin era, today, the number of Github and bitcoin related projects more than 10000. As the 2 version of the block chain, Vitalik Buterin launched the Ethereum project in 2013 and released the first version of the "Ethernet Fang white book" in December. Finally, Wang Wei indicated that we should call for a joint participation in the big sea age of block chain.

  • 9:00[big data core technology and application practice summit], under the auspices of the Chinese Academy of Sciences Research Institute of computing technology under the auspices of the ceremony, the official start.

Sina micro-blog (#CCTC#)@CSDN cloud computing

Guest speaker

At Keynote am on May 18th

CSDN, geeks help VC, AI100 founder Jiang Tao
Open source China CEO Ma Yue
Mesosphere co founder and CTO Tobias Gunter Knaup
Primeton CTO Jiao Lieyan
Rancher Labs co founder and CEO Liang Sheng
UCloud block storage R & D vice director Peng Jingxin
Senior director of Jingdong group's operations research and development department, Li Pengtao

Micro service special

Chairman of the 58 group technical committee Sun Xuan
DaoCloud chief architect Wang Tianqing
Financial data Beijing R & D center CTO Wang Dong
Mesosphere distributed systems engineer Gilbert Song

Cloud computing core technology and Practice

Special session of artificial intelligence

Big data core technology and application combat Summit

Vice president of Institute of computing technology, Chinese Academy of Sciences
East Guoxin large data center director Wang Hu
Chen Kang, associate professor, Department of computer science, Tsinghua University
Hortonworks technical expert Liang Yanbo
Beijing even Technology Co., Ltd. architect Tao Zhenglin
Apache HAWQ founder Chang Lei
Beiyou computer department associate professor Xiong Yongping
Didi travel Senior Software Development Engineer Li Yang
Kim Union Bank chief technology officer shangf
HanSight, senior technical engineer, Zhang Jun
IFLYTEK real time computing platform senior architect Liu Jun

Container and Operations Technology Summit

Spark Technology Summit

Block chain technology summit

Shanghai Le live CTO Wang Wei
People Insurance Information Technology Engineer Li He
Distributed technology CTO& founder Zhang Zhengwen
Chi chain technology CEO Dong Ning
Round table dialogue
Bit founder, "talk about block chain" tutorial author Tan Guopeng
IBM Greater China Laboratory Services Division Chief Architect Li Yansheng
Haina cloud CTO Zou Jun
Shenzhen silver chain technology, CEO, gold chain, deputy secretary general, Shen Tu youth
Founder of integral coin Zhang Lei
Round table conference on how to learn block chaining
Mesosphere co founder and CTO Tobias Gunter Knaup
Primeton CTO Jiao Lieyan
UCloud block storage R & D vice director Peng Jingxin
Rancher Labs co founder and CEO Liang Sheng
IBM GBS Watson senior consultant Jin Jie

Conference highlights

Attendees in line at the sign in
Bustling exhibition area
CSDN I art is on the air
On a hot summer day, have a cool ice cream
Beautiful foreign models interact with the audience
All seats are occupied. venue
Dry too much, have to take PPT learning
Technical communication, across the language
Delicious tea to fast start
I love CSDN

Schedule at a Glance