Shulou(Shulou.com)06/03 Report--

1. Cloud computing philosophy (organizing resources for services, organizing technologies for implementation, organizational processes for adaptation): services and platforms

(1) definition: refers to both the provision of applications in the form of services on the Internet and the software that provides these services in the data center.

Cloud: hardware and software in data center

(2) predecessor: very large-scale distributed computing

(3) Core idea: general purpose, dynamic:, multi-lease

Flexibility: control loss (group communication: single point of failure, redundancy mechanism, multi-live mechanism, message sequence mechanism),

State awareness (member management services, global views),

Dynamic balancing (dynamic balancing of load by increasing the number of nodes, DHT algorithm)

Transparency: redundancy and other ways to ensure that the underlying architecture of the system node problems, will not affect the whole

Users do not need to care about the underlying implementation, as long as focus on the upper business logic

Self-fault tolerance (session replication technology, node memory connection to global shared memory data space),

Self-management (distributed execution mechanism, policy-based automation management, event-driven mechanism),

Cross-platform interoperability (universal for users, common for applications, common for platform layer)

Building blocks: business process management BPM, rapid resource reorganization to meet different business needs

(4) Model: business model (cloud service), technology (cloud computing platform that provides powerful resources on demand).

(5) challenges: big users, big data, big systems

(6) limitations: bandwidth, security, atomic operation (minimum indivisible) delay

2. August 2006 SES search engine strategy Google CEO Schmidt named cloud computing

Amazon EC2 (dynamic Computing Cloud) Service defines Cloud Computing Business Model in 2006

3. Cloud business model:

1) SaaS Software as a Service (Internet provides software services): provide services on demand, no need to purchase software, membership fees, no management and control of the underlying cloud infrastructure

2) PaaS platform as a Service (Internet provides application running platform services): delivers computing platforms and solution packs in the form of services, providing a highly integrated environment for creation, application testing and application deployment without the need to purchase hardware and software

(1) do not manage and control the underlying infrastructure: network, server, operating system, storage, etc.

(2) the deployed application can be controlled, and the service provider provides the application service programming interface.

(3) eg: Google GAE develops hosted network application platform

3) IaaS infrastructure implementation as a service: the Internet provides infrastructure services

(1) consumers have access to processing power, storage, networking and other basic computing resources

(2) can deploy and run any software such as operating system and application

(3) website hosting, data storage and backup, e-commerce, high-performance computing, search engine, virtual hosting service, etc.

(4) eg: Amazon AWS (storage, computing, messaging (SQS) and data acquisition (SimpleDB) services)

4. There is no obvious difference between SaaS service-oriented object and ordinary stand-alone application customers.

PaaS users are developers and need to understand the development and deployment of applications in the environment provided by the platform.

IaaS's lowest-level IT infrastructure service for IT managers

5. Deployment usage

1) Public cloud Public Cloud: cloud infrastructure open to the public

2) Private cloud Private Cloud: a cloud infrastructure built by customers alone to provide the most effective control over data, security, and service quality, and to control how applications are deployed on this infrastructure

. [difference between public cloud and private cloud] is mainly reflected in commercial applications, technical interoperability, but fundamentally different access permissions and access mode control at the technical level (the scope of access determines the business nature of the cloud)

3) Hybrid cloud Hybrid Cloud: semi-public and semi-private, resource borrowing / leasing

6. Cloud computing technology genre

Yahu: search start, Portal Mode

The most consideration for the development of large-scale distributed systems is versatility (supporting a variety of applications, non-storage problems).

Unresolved mailbox storage and search efficiency issues to drive Hadoop

Google: search starts to ensure the efficiency and availability of query and search for large chunks of data

Storage as the root, cost, high efficiency and new availability

Amazon: e-commerce, the core role of the underlying architecture to ensure uninterrupted online transactions

Dynamo is more suitable for storing relatively small objects.

7.Hadoop: a software framework for distributed processing of a large amount of data, which is mainly composed of HDFS (file system), MapReduce (computer system), Hbase (database), etc.

Driven by Yahu, technical ideas belong to Google

8. GFS, HDFS: one master server (Master) + multiple child table servers (Chunk Server)

9. Dynamo: the hash algorithm is used to split the data, and the data is stored evenly in a single "ring" to share the pressure.

10.VMware: mainly provides virtualization technology to enterprise users, and determines that its technology path is to deploy cloud computing platform from standalone virtualization, upward, to a higher level of development.

Features: "pooling" of various hardware resources realized by virtualization technology, on the basis of night configuration, scheduling and management

11. Distributed system: connecting physically dispersed computing resources through the network to solve problems

twelve。 Stand and say it. Meaning of data volume: total amount, flow (concurrent amount)

13.cebook stores data on Hadoop distributed file system (HDFS) and relies on Hive to complete data analysis.

14. The system consists of many cheap common components, and component failure is the norm. The system must monitor its own status at all times and quickly detect, bear and recover failed components.

15. Factors for the rapid development of cloud computing

1) Technology: optical fiber use, broadband popularity, node communication, virtualization technology, Ajax, LAMP

2) Business model & requirements: software as a service

JavaScript and XML Technology of 16.Ajax:Asynchronous JavaScript XML Asynchronous

The form feedback from the server to the client will not refresh all the page content because of some changes, but only need to return the necessary data for the change.

Everything can be served by 17.XaaS. Derived business processes as a service, database as a service, security as a service, etc.

The service can be accessed at any time, the service should always be online, and the service has a large enough user base.

18. Parallel computing: a computing mode in which multiple instructions are executed at the same time to speed up computing. Tight coupling

Coupling: the dependence of systems that interact with each other

Distributed computing: loosely coupled, connecting distributed computing resources in a transparent and scalable way to solve problems

[comparison] (1) the resources of distributed computing are scattered, showing a shared architecture through network node connections.

Each processor has its own memory, which exchanges information through message transmission between processors.

(2) parallel computing is usually concentrated in one place, and information is exchanged through shared storage.

All processors share one memory and exchange inter-processing information at once

(3) NUMA (non-uniform storage access structure) and MPP (massively parallel processing) each node has its own CPU, memory and I-bank O, and the nodes interact with each other through certain mechanisms.

19. Cluster: using multiple computers and redundant interconnection to form a single high availability system for users

20. Network computing technology: connect the computer nodes scattered in the network (such as idle servers, storage systems, etc.) together to form a virtual computer with super performance, providing users with powerful computing and storage capacity.

Loosely coupled and heterogeneous properties are more obvious.

21. Grid computing: multiple fragmented resources provide a running environment for individual tasks

Cloud computing: a single integrated resource provides services for multiple users

22..SOA (Service Oriented Architecture) a service-oriented architecture that organizes and leverages decentralized functions that may be under the control of different ownership scopes.

Given a quasi-interface on the other side and a service protocol with a constraint excuse, any business application can communicate and interact through a given standard interface as long as it is satisfied.

Essence: enterprise integration technology for exchanging information between systems

23.EAI (EnterpriseApplication Intergration) Enterprise Application Integration Technology: message-based Middleware products

24. Virtulization: logical representation of resources without physical constraints

Abstract the underlying resources, provide the same or similar functions as the real layer, and shield the difference and compatibility of devices.

25. Cloud computing load balancing methods: horizontal segmentation (according to workload), vertical segmentation (according to workflow)

twenty-six。 Business application architecture: presentation layer, business layer, data layer

twenty-seven。 Distributed data bus: operation and maintenance management system, data acquisition system, data storage system, data service system, service delivery system, distributed scheduling system / distributed workflow engine

(1) distributed scheduling system / distributed workflow engine: system executor, implementing definition, scheduling, collaboration and execution, mainly supporting the automatic operation and maintenance of large-scale systems and the implementation of distributed computing framework

(2) Operation and maintenance management system: configuration control. Rights management, monitoring and alarm, application management, fault management, integrated policy engine, distributed workflow, resource and service management scheduling, etc.

(3) data acquisition system (starting point of data-driven architecture): configuration management, data filtering, data preprocessing

(4) data storage system (any information system core): unified storage system

Relational database cluster: storage and fast access of massive strong relational structured data

Distributed real-time database: the amount of data such as user behavior data and system log data is large, and the correlation is not strong.

Distributed file system: all kinds of pictures, videos and other media files with large total data and total capacity

Real-time and efficient storage and access of unstructured data

(5) data analysis system: distributed computing model (data transformation, data aggregation, data association, data mining)

(6) data service system: the bridge between the supporting platform of enterprise information system and the upper business application.

Service management system: service registration, logout, change, audit, release, design

Data access service-data cache: database access, real-time access, file system access.

Business logic service-data cache: push storage system, intelligent analysis, statistical reports.

(7) Service delivery framework: front-end server, delivery server, load balancing

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.