First, the commercial product technology
Behind TOP1 286.4 billion turnover Ali cloud storage system Pangu 2.0 rock solid
2019 global day cat shopping carnival hit 286.4 billion of commercial miracle, as a cloud infrastructure stability and performance engines, Ali cloud self-development of a distributed storage system Pangu 2.0 provide services rock solid reliability. With ESSD cloud storage disk block large-scale applications, Pangu 2.0 successfully support the database, middleware, MetaQ, Sigma online services, ants gold dress, and other transactions rookie network link, enabling Ali economy double eleven core system 100% the cloud.
TOP2 operating system, JVM, compiler products support cloud scene on important core group to build a solid two-eleven, advanced infrastructure
kernel level to solve the cloud BLINK, PAI, container platform mix and other business critical technological barriers, and for the first time using the program on a NUMA million units scale servers, and by quantifying the SPEED platform analysis demonstrated significant performance improvement program, built on proven technology evolution native community infrastructure software is successful. Star product ali-diagnose solve a series of major issues crucial pressure test to ensure a stable and efficient zero big promotion. JVM statically compiled timeout applied to achieve a cloud application startup speed. Kangaroo secure container for the first time on-line double 11 support a hybrid deployment PAI and ASI, in order to follow a comprehensive upgrade to lay the foundation for the program based on the new Department of mixed kangaroo.
TOP3 proprietary network VPC supports single-tenant instance of scale in the world, the perfect support group to promote dual 11 large core systems on all public cloud
2019 Group Alibaba promote dual 11 large core systems on all public cloud, which requires the construction of large public cloud the size of the virtual cloud data center. VPC team another way by mixing the Ministry memory sharing technology, routing relay technology to solve the large capacity vpc memory footprint is too large, routing updates slow convergence industry problems, breaking many restrictions, to build the world's largest VPC network on public clouds, steady support double carrier 11 vpc hundreds of thousands of single container instance demanding requirements, provides a simple and reliable, low latency, high throughput network cloud double eleven electricity supplier and payment core business.
- Ali cloud released hybrid cloud CPFS one machine, as customers embrace cloud transformation of traditional industries to provide the best tool
December 2, at the Guangzhou summit, Ali cloud officially released hybrid cloud CPFS one machine, it can support the largest single 2.6GB / s throughput capacity, expandable single cluster nodes 9620, hundreds of GB / s throughput, to provide distributed flash acceleration data-intensive. Meanwhile, hybrid cloud CPFS one support multiple copies, the erasure mode data redundancy codes, customer data to provide protection for the ultra-high reliability, can be quickly detected and the failed disk serving node down to achieve second fault detection.
The storage capacity of the unit, the official beta, the first general store selling shape the world premiere of
storage capacity per unit packages (Storage Capacity Unit, hereinafter referred to as SCU) formally launched the beta, it can be used to offset the many different types of cloud disk according to the amount of pay the bill. Compared to the cloud server with the purchase of pre-paid mode, SCU combined with the amount of pay-per-use cloud disk, both flexibility and cost-effective use of resources.
- Snapshots service upgrades - create and retain up to 1000 single-disc automatic snapshots
A snapshot is a point in time a cloud disk or a shared block storage data state file. Commonly used in data backup, data recovery, making custom images, applications and other disaster recovery scenario. In order to meet more stringent data compliance requirements and provide more secure data protection services, Ali cloud snapshot service has undergone major upgrading, single cloud disk can be created and retained 1256 snapshot, which includes 256 manual snapshots and 1000 automatic snapshots .
Document links: https://help.aliyun.com/document_detail/25391.html?spm=a2c4g.11186623.6.797.307f341fv4mGxD
- SLS and Ali cloud cost to build center launched the "cost butler" APP, to solve complex cost structure behind the clouds on the face of corporate customers, the cost is difficult to predict the universality of
SLS and Ali cloud launch cost centers based on existing functionality and build log service "cost butler" APP. APP housekeeper cost data acquisition, analysis, forecasting, early warning and the ability to visualize the report, presented to customers can pre-customized reports, and support for SQL-based interactive analysis capabilities. Help solve complex cost structure behind the clouds on the face of corporate customers, the cost is difficult to predict the universality of the problem. - ALIYUN :: ROS :: CustomResource release
can support non Ali cloud resources, such as AWS EC2, meet GAP support hybrid cloud.
Operation and maintenance scheduling feature upgrades, the official release trigger an alarm, set alarm thresholds by monitoring cloud-term monitoring of the implementation of predefined templates triggered by an alarm, to automate the operation and maintenance operation, to achieve self-healing effect, such as instances when ECS when the cpu utilization exceeds 90%, an alarm is triggered, automatically perform load balancing solution from the tie, then mount an operation to repair an example.
Choreography operation and maintenance experience to upgrade the line batch operation example fast page on the console, for more than one instance of script commands or batch execution scenes start and stop operation, no remote connection examples and templates written in script commands to be executed after the console input , you can quickly select instances batch operations, and support tag filtering, concurrency and lots of customization options strategies to further reduce the threshold for automatic operation and maintenance.
Elastic container instance ECI support and resource group created out of account. It applies to companies split and sub-business resources based on business management and control department.
Specifying Snapshot disk size elastic properties of mirrored cache container instance ECI official Release, support ImageCache create mirrored cache, support expires deleted.
ECI elastic container instance to publish user-created prometheus, self-built support log system.
- Ali cloud microphysical Service Engine MSE 1.4.0 version of the blockbuster released
on the basis of the existing ZooKeeper engine, adds support for Nacos and Eureka, as well as in monitoring trends to add a custom alarm notification function monitoring indicators, notification method support SMS, e-mail and nails robot, covering Shanghai, Hangzhou, Beijing and Zhangjiakou region. Details: https://yq.aliyun.com/articles/727791 - Ali cloud Serverless Application Engine (SAE) released v1.2.0, savings of up to 57% of the unused computing resources
Ali's first cloud-oriented applications Serverless PaaS products - Serverless Application Engine (SAE) heavy upgrade, support for a key start and stop, NAS storage , practical features such as small size instances. Details: https://yq.aliyun.com/articles/727869 - Distributed enterprise application services EDAS upgrade micro service governance capacity, introduce outlier removal feature examples
of enterprise applications distributed services EDAS upgrade micro service governance capacity, introduce outlier instances removal function, which can detect the Spring Cloud and application examples of Dubbo availability and dynamically adjusted to ensure the success of the service call, so as to enhance the stability and quality of the services business. Details: https://help.aliyun.com/knowledge_detail/145690.html - Real-time monitoring service ARMS release a new version
of the version completed X-Trace and ARMS chain of calls get through, support Agent version / front-end version of the gray-publishing, stripping the products depend on Dauth, the on-line support applications grouping alarm function, micro-services OneAgent elegant offline support officially released, Prometheus monitoring of new Flink / ApiServer such as default support the broader market. - VPC whole new generation of hardware gateway support group Dual 11 big promotion, business networks silky smooth flow of flood peak at
2019 Dual 11 big promotion group Alibaba core systems on all public cloud, the traditional X86 architecture has failed to meet such a large bandwidth requirements, in response to soaring peak traffic, protect the stability of the two-eleven, cloud network team focus on investment XGW2.0 gateway hardware and software research and development, the Group of 11 double Zhangbei, Shanghai, Shenzhen region three public network and private line services full switch to XGW2. 0 architecture. During double eleven, XGW2.0 cluster rock solid, placid, business cloud silky smooth. - Network products support the full build ARM-based servers, to meet customer demand for a specific scene
networking products VPC, SLB, and all other products to fully complete the localization of ARM adaptation work to complete the POC in NU20 environment, and pioneered the use of proprietary clouds, to meet specific customer there is a scene specific needs of the ARM server. - SLB on-line across the region to mount RS and other features on the new line to meet customer multi-application scenarios
SLB recently released a number of functions, SLB For CEN function, greatly enriching the user's network scenario, the scheduling inlet flow to different regions of SLB on the ECS, improve inter-cloud network traffic scheduling; a key replacement certificate function, pain points on the operation and maintenance of a large number of repeat work to resolve the big customers to replace certificates, as well as drain brush certificate; the new health check function to solve the old version health checks a user can not view the forwarding rules, check the status of RS health problems, users can more clearly observed their business operation. - ACK container service expansion capabilities Enhanced Cluster, can be used to support multi-region expansion, support for multiple data disc mount
ACK continued to improve cluster expansion capability, and create a consistent cluster, now the user can select a plurality of available area for expansion. Further, the expansion can be mounted a plurality of node data disc while the disc is set to user data may be encrypted disk. - Container Service ACK node capabilities Enhanced Cluster, support for custom scripts, Tag, OOS
ACK enhanced cluster node capabilities, users can create and node add custom scripts (userdata) (whitelist open), the ability to customize respect to the expansion of the cluster Some users need a particular node of OS is important, users do not need to package custom images, but directly to the standard script injected into the mirror, greatly enhancing the flexibility. Node tag for the user to provide a convenient split resources on the node, the node function supports auto-scaling function. OOS is choreography operation and maintenance services, joined the ACK node in the node OOS docking maintenance functions, the user can jump from ACK to the interface OOS, OOS for the ACK node performs operation and maintenance script. - ACK Serverless Kubernetes container services available to support multi-zone, log audit, unified CCM
Serverless Kubernetes architecture since the 2.0 update, continue to strengthen the consistency of native cloud experience. This month more than the required new functionality available area, log audit function, creating service when CCM control components and also to maintain a consistent standard Kubernetes. Kubernetes competency standards will be further reflected in the Serverless Kubernetes. - Container Service ACK cluster support vGPU resources
ACK enhance the ability of AI, the big data computing, public cloud vGPU docking infrastructure resources (vgn5i). For now, you can create a container cluster vgn5i type of the resource. - Container Service ACK cloud-native network Terway support ENI cache
Terway is based container network plug Ali cloud ENI technology, this feature allows Terway creates a ENI buffer pool when the cluster node initialization, which initialize certain number of ENI IP, which will to some extent accelerate the creation of the Pod, enhance the user experience. - Container Service ACK CCM enables users to mount ECS SLB back-end
CCM Management Service is mounted to the SLB system components, general Service cluster node where the SLB will mount to the back-end. This capability allows the user node outside the cluster SLB access to the rear end, the container application Service shared access to external traffic, in which the user application migration stock, grayscale, etc. scene very useful. - Proprietary cloud security V3.10 release SOC, cloud firewall, network traffic monitoring product
1, a proprietary cloud security SOC: providing a global platform for tenant security and unified security management capabilities and API, to meet the large-scale private cloud customer focus groups and industry safe operation and the like Paul 2.0 centralized control requirements.
2, exclusive to North Yunnan Firewall: providing a private cloud to the north and south boundaries of the original biochemical isolation control and protection, border security does not allow private cloud streaking.
3, a proprietary cloud network traffic audit: Dedicated private cloud network traffic monitoring (CBeaver) released a comprehensive collection depth analysis of private cloud CSW side flow into and out of ten yuan group and DPI, abnormal traffic and sophisticated attacks, and defense and evidence to deal with the Internet and intranet boundary risk control requirements.
- Access Control (RAM) supports user credentials report and AK last time query functions
access control (RAM) is to provide Ali cloud of managing user identity and access to resources and services. RAM introduced a new reporting user credentials, you can log in the console RAM ( https://ram.console.aliyun.com/ ), click on the "Download user credentials Report" in the overview page for a csv file format, which lists your cloud account and all the user's login credentials RAM information, including console login password, access key (AccessKey) and multi-factor authentication. You can download the user credential report regularly to enable the user to log audit Ali cloud conditions. In addition, the detail page for each user, if the user creates access key, you can see the last time the access key to help you confirm that the use of access keys to avoid accidentally deleted is in use access key. Document links: https://help.aliyun.com/document_detail/143477.htm
Second, the technical progress of the project
TOP1 ECS local SSD disk Example: stand-alone exceed 500 million performance
ECS local instance of SSD disk resources help OceanBase summit TPC-C! Ali cloud launch of open test platform, OceanBase database server using 204 + three models are ecs.i2.16xlarge Ali cloud ECS server, where 204 units as datanode, as well as the three root node. Under NVMeSSD highest performance and enterprise-class cloud Ali optimize read and write performance of the whole link, I2 IO provides ultra-high storage capacity, ultra-low microsecond (us) level access latency, high-level mega + IOPS, 10 + GBps large throughput level, to bring the database faster response speed and processing power, corresponding to the specification examples is "SSD type local specifications group example i1, i2, i2. team memory block by multiple queues, mempool , nvme io interfaces, driver and other aspects of technology research and ultimately the extreme IO performance
TOP2 Goddess Pooling / services in the iterative development of
front-end technology: shmq CPU consumption is slightly higher at sysbench 256 thread OLTP model, continue to improve the polling mechanism to reduce idling
back-end cluster: complete recovery framework merging into the main line; perfect orphan file space reclamation ; complete disk heartbeat detection scheme design and optimization
DBFS Master: increased deliveries tenants, delete and display function; QoS metrics designed to facilitate monitoring and display Shennong
TOP3 complete StackGroup design & review
into the technical evolution stage, enabling users across the cross-domain deployment and management capabilities are expected to be on line by the end of 12
Third, academic, scientific, and technological achievements
TOP1 Ali cloud-based software as a corporate representative to the CCF committee for social services and to enhance the contribution of Ali cloud influence
November 22, 2019, CCF system software professional committee annual meeting in Hangzhou, infrastructure software, senior technical experts hope CCF primary election system software IPCC Standing Committee, also serves as CCF software engineering IPCC Standing Committee. Ali cloud-based software services continued to contribute to society, and to raise its profile in the academic field.
TOP2 Alibaba again OpenJDK Committer get a seat
at Alibaba, AZul, Redhat, Datadog jointly promote, JFR technology is expected to enter the standard distribution version of Java early next year. As cloud computing era curtain went up, OpenJDK community to be involved in more and more companies.
TOP3 OpenMessaging won the eighteenth Japan and South Korea in Northeast Asia excellent open source project
OpenMessaging project initiated by Alibaba, and Yahoo, and pieces of travel, Streamlio company to participate in the creation, the project is intended to the creation of vendor-independent, platform-independent and distributed message stream processing standards in the field of application development. Has experienced several years of development, OpenMessaging absorption including Yahoo Japan, EMQ added, so far the world has more than a dozen open source projects with the cloud vendor members to join, won the eighteenth Japan and South Korea in Northeast Asia excellent open source project
- Ali goes through the ISO / IEC 27701: 2019 standard certification scored privacy certification "Grand Slam"
Recently, Ali cloud obtained from well-known international certification body BSI standards issued by ISO / IEC 27701: 2019 privacy protection certification, which is the industry's most authoritative privacy management system construction guidelines, once again reflects the commitment to adhere to Ali cloud user data privacy protection. Up to now, Ali cloud has received include ISO / IEC 27701: 2019, ISO / IEC 29151: 2017, ISO / IEC 27018: 2014, BS 10012: 2017 , including all about privacy protection standards certified "Grand Slam." - The number of "CNCF x Alibaba native cloud technology open class" people always learn 14009
- CCF IPCC, young scholars in the field of storage to team talks and exchanges
to enhance industry-academia cooperation between the team and the professional committee of CCF, improve storage team to the forefront of research and understanding of academic exchanges, invited CCF IPCC No. 11.29, storage young scholar in the field to team talks and exchanges. About 90+ students participated in the storage team sharing sessions and a lively questions and communicate lectures.
No. 11.29 in the afternoon, Professor Lu travel CCF Youth IPCC Tsinghua University, Professor Li Yongkun China University of Science and Technology, Professor Patrick PC Lee Chinese University of Hong Kong and Harbin Institute of Technology Professor Xia and storage teams Dongping, Wen Tsui, Hui Pa, Chu Road, Lee Fei, Yang Shan, Wang Yong, Mansfield and other students of communication problems in the enterprise and cutting-edge technology research hot topics expected to explore new modes of cooperation and student training methods on industry-academia cooperation.
- Fast storage team technological achievements
Citrix (block an efficient method of upgrading heat storage virtualization end access module); Ling a (Heterogeneous data storage system allocation algorithm);
(Single disk level cloud intelligent flow control device based on user information) is sharpened; Zhang Hang (a novel message queue flow control means)
- Ali cloud object storage service OSS won the "2019 Year Buy-Award" in the "cloud storage software products Gold Award 2019"
- Ali cloud hybrid cloud backup service HBR won the "2019 Year Buy-Award" in the "2019 Annual Data Protection Product Innovation Award"
Fourth, Ali cloud preacher style
TOP1 Ali cloud preacher project leader Hoan Kiem published "Ali cloud ToB genes from where"
in the well-known new media platforms new artificial intelligence Ji-won, Hoan Kiem published "Ali cloud ToB genes from where" heavy articles to plain language and clear logic to explain the birth of Ali Baba 20 years of accumulated ToB ability, contribute to the campaign ToB Ali cloud.
TOP2 Guido Live "Ali cloud open platform: Why? What? How? "
Ali cloud open platform Guido, head of the group at the same time live in a plurality of staples such as Ali cloud evangelist" Ali cloud open platform: Why? What? How? "Ali cloud direction for the three open platform: the enterprise IT governance, open API, corporate table, a comprehensive introduction to the status and direction of development of an open platform.
(Finish)