solution
Home > solution > Cloud Computing Products

Cloud computing solution of HWTT

Overview of the plan

With development of internet technology, more and more application is cloud-computing oriented. The cloud computing is the product arising out of development of traditional computers and network technology including network computing, distributed computing, parallel computing, utility computing, network storage, virtualization and load balancing. The core idea of cloud computing is to carry out unified management and scheduling of mass computing resources connected via networks, so as to form a computing resource pool for providing on-demand service for users. Cloud computing is the change of the method for providing IT service. It is expected to enhance efficiency and flexibility under the circumstance of geometric growth of the data center. A lot of cloud computing deployment relies on computer cluster, including computing, network connection and storage. The chart shows the topology deployment for specific cloud computing application.

3 the topology chart 3 solution

Chart 4 scheme 3 structure

The aforesaid chart describes the cloud computing application. Its main business application supports Oracle Rac, distributed communication record analysis and etc., in the operation and analysis system. The cluster server nodes are equipped with 32 blade servers. Each node is equipped with 40Gb InfiniBand QDR HCA network cards which are connected to MIS5030 IB switches. Many business application servers are connected to core 10 GE switches. The Ethernet and IB network are connected via MBX 5020. MBX 5020 has four IB ports. Each IB can be connected to 3 GE ports, thus realizing communication between computing nodes and Ethernet servers.

  ◆VSA server

VSA is storage acceleration software which is equipped with two servers. Each server is equipped with one HCA (IB) card, two HBA (FC) cards and one 512GB SSD. The VSA server serves as the gateway equipment, realizing transformation from IB to FC.

  ◆HCA card

40Gb InfiniBand QDR HCA network card.

  ◆HBA card

8Gb FC card

  ◆SSD

Use electronic hard disk products of CFD-SATAE with continuous read-write capacity reaching 200 MB/s, for loading VSA software and serving as a cache.

VSA is connected to FC switch and back-end FC storage.

In the aforesaid topological graph, integration and intercommunication between IB, Ethernet and FC network are realized. Such application is used for network deployment of different cloud computing requirements.

For current enterprises, 80% of data are unstructured. Such data increase by 60% annually according to indexes. Big data will become a test of the storage architecture of the enterprise and basic facilities of the data center and etc., and may lead to a chain reaction of application of data warehouse, data mining, business intelligence and cloud computing and etc. In the future, enterprises will use dataset of over 1 TB for business intelligence and commercial analysis. By 2020, the volume of global data use is expected to increase by 45 to 35.2ZB (1ZB = 109 TB).

Distributed computing should be used for supporting analysis of such data, and thus mining their intrinsic values. The previous log analysis design of multi-thread and multi-task decomposition is the simplified stand-alone edition of distributed computing. Distributed computing framework design involves how to divide and turn such stand-alone work into clusters of collaborative work.

Application of the distributed computing to multi-scenario and mass-data distributed processing is required in the business contents of cloud computing service. InfiniBand is a type of network for fields where there is high requirement of bandwidth delay. At the current stage, the mainstream bandwidth is 40Gb/s. The end-to-end delay in the network can be controlled at the US level. InfiniBand network features centralized management and support network division, QOS and other functions, and is convenient in expansion, and can have over thousands of nodes. It is approved that InfiniBand network is applicable to application of MPI, ORACLE RAC, HADOOP, and etc.

网站备案号:京ICP备15015354号-1 京公网安备 1101 0802 017216号