Analysis of the cloud-based railway data center

Under the background of the information age, after years of development, railway informatization has become an important support for railway daily work. In the transportation organization, the passenger and freight marketing, the management and the safety monitoring and so on work, each kind of information system plays the important role to become more and more manifest. Therefore, to meet business needs, in the Transport Depot section, the Railway Bureau, the Ministry of Railways has established a variety of levels of information systems room, and according to the business maximum data and calculation requirements of the acquisition of host, storage equipment and the corresponding platform software, in order to ensure the safe and reliable operation of information systems, most of the main system using the main Single (or multiple) storage array structure. These equipment are distributed in the Railway Bureau or station section room. In operational practice, the following problems exist: (1) The infrastructure is huge, and the investment and maintenance costs are increasing. (2) The use of resources is unbalanced and the utilization of resources is low.

With the speed of railway, traffic volume and quality of service requirements continue to improve. There is a need to study new information systems infrastructure, integrate existing resources, build cloud-based data centers, meet new challenges, and reduce the rising cost of information systems investment and maintenance.

1. Cloud Juice Count

It is generally believed that cloud computing is a network resource sharing mode that accesses the computer and storage system according to requirements by making the computation distributed on a large number of distributed computers or remote servers, according to the mode of Internet operation, and switching resources to the required applications. In this shared mode,. Cloud refers to a variety of virtualized computing resource pools that include the infrastructure for building applications and the specific cloud computing applications on these infrastructures.

Cloud computing architecture as a shared infrastructure approach, although there are different formulations, but in essence, is the dynamic distribution of computing on the cluster computer, according to the business development and application of software and hardware resources to dynamically deploy, to achieve on-demand access, and provide a variety of IT services internally and externally. Cloud computing is developed by technologies such as distributed computing, Grid computing, and integration of technologies such as virtualization and SOA.

According to the cloud computing operation mode, it can be divided into private cloud and public cloud. Public cloud computing is deployed in the service provider's environment, providing computing services for different customers at the same time. Private cloud computing is deployed in the user environment, and only provides computing storage services for users, enterprises have the ownership of resource allocation, and can improve their services based on their own needs, and carry out independent innovation. Large enterprises can develop their own private cloud computing, enjoy the benefits of high scalability, easy maintenance and access to cloud computing, and secure data.

Cloud computing systems can be categorized according to the level of service they provide, mainly providing infrastructure tier services, providing platform-level services, and providing application-layer services.

(1) Infrastructure layer services. Infrastructure layer mainly includes computing resources and storage resources, the entire infrastructure can also serve as a service to users, not only the virtual computing resources and storage, but also to ensure the user access to the network bandwidth. This is also the model of cloud computing recommended by hardware product providers.

(2) Platform layer service. The platform layer on the infrastructure includes parallel programming and development environment, distributed storage management system with massive data, large data Distributed File system and other system management tools for cloud computing, such as deployment, distribution, monitoring and management of resources in cloud computing system, Ann

Full management and distributed concurrency control. Platform layers are designed primarily for application developers, and developers do not have to worry about the resources needed to run the application, and the platform layer provides all the platform resources needed to run and maintain the application. Googleappengine Cloud computing Services, Amazon's cloud computing platform is such a platform. Platform Layer Service technology threshold is relatively high, still in the continuous development.

(3) Application layer service. It is to provide users with a simple software application services and user interface, and so on, users do not need to purchase and manage server-side software, known as software as a service. For the traditional software, the Application Layer service solution has obvious advantages, including lower upfront cost, easy maintenance, rapid deployment and so on. The underlying layer of the application-tier service can be based on the infrastructure layer or platform layer.

2. Data center of railway Information system based on cloud computing

2.1 Infrastructure Layer Integration resources

With the integration of resources at the infrastructure level, existing application system software can be simply ported to new environments. Virtualization is a fundamental design technique for cloud architectures that allows servers, storage devices, and other hardware to be treated as a pool of resources, rather than as discrete systems, and can be allocated as needed to quickly deploy and increase workloads by providing virtual machines or physical machines quickly. In addition to server and storage consolidation, it provides an integrated system architecture, including application infrastructure, data and databases, interfaces, networks, desktop systems, and even business processes.

At present, the main engine of railway information system includes two kinds of small (medium) and microcomputer server, in the current technical conditions, minicomputer and micro-computer server virtualization is different, minicomputer such as IBMP595 only on a single machine to complete the aggregation and disaggregation of resources, not across multiple IBMP595 to complete such resource integration, There is no uniform standard for various types of minicomputer virtualization, which can be integrated using the virtualization and cluster technologies provided by the manufacturer based on the availability of minicomputer. and the computer server in the virtualization aspect is more mature, support across the vendor, cross model of virtualization integration. The computer server should use high-end mature products as far as possible, and the low-end servers that are insured should be discontinued when considering total cost of ownership. The original resource allocation object is a microcomputer server, which needs to be installed and deployed after the realization of computer server virtualization, the object is the virtual machine, can be directly implemented in the deployment process of the system quickly installed and deployed. Through the virtualization technology, the computer server resources owned by the enterprise are pooled and shared, and the idle resources can be fully utilized through the sharing and dynamic allocation of IT resources, thus increasing the utilization of resources. Before virtualization, the computer server and storage utilization of enterprise data center average less than 50% (the actual utilization rate is usually lo%~15%, one is prepared). With virtualization, you can transfer workload encapsulation to idle or less-used systems, which can consolidate existing systems, thereby delaying or avoiding the purchase of additional server capacity.

Based on the virtualization of resource integration, especially after the virtual computer server, can greatly improve the reliability of resources, by reducing the number of servers can reduce power, energy consumption and data center room space. More applications, including previously used minicomputer projects, can be ported to virtualized resource pools to reduce investment and maintenance costs and improve operational reliability. Of course, it is very important to manage these resources, which can cause huge losses.

2.2 Platform Layer Integration resources

The integration of resources in the platform layer requires the establishment of a new software platform, a relatively high technical threshold, the existing application system transplant is more difficult, in the development of new applications or upgrade existing systems can be used in this way. IBM's Blue Cloud computing platform is a sales computing platform that allows users to build their own cloud platforms based on these hardware and software products. The Google Cloud computing platform is based on a complete set of distributed parallel clustering infrastructures. Other manufacturers also have a number of unique computing platform. The following is a combination of Google's cloud computing platform to illustrate the focus of consolidating resources at the platform level.

2.2.1 Distributed File System

The Distributed file system has the characteristics of scalability, reliability and usability, which splits the data file into different blocks or as a complete block of data, which is stored as the basic unit and stored in the file data Block node. For disaster-tolerant purposes, each block of data is stored in a different data node.

Google's cloud computing platform employs its unique distributed file System GFs, where each block of data is stored at least 3 different data nodes, deploying many GFS clusters, some with more than l000 storage nodes, and more than 300T of hard disk space, Continuously accessed by hundreds of clients on different machines.

2.2.2 Distributed Database

Distributed database system includes distributed database management system and distributed database. In a distributed database system, an application can transparently manipulate a database. A distributed database is logically a unified whole and is physically stored on separate physical nodes. An application can access a geographically dispersed database through a network connection. The distribution of the data in the database is not stored in the same site. Rather, it is not stored on the same computer's storage device. The Google Cloud computing platform employs a large distributed database management system (BigTable) that can handle both formatted and unformatted data.

2.2.3 Distributed programming

The distributed model calculation process is to decompose the data set to be processed into a small dataset, and the dataset is distributed in a node of a cluster to process and generate intermediate results. These intermediate results are merged by a large number of nodes to form the final result. Google constructs Map-reduce programming specifications to simplify the programming of distributed systems. Application writers simply focus on the application itself, while the handling of the cluster, including reliability and scalability, is referred to the platform for processing. MapReduce uses two simple concepts such as "map" and "reduce" to form the basic unit of operations, and users simply provide their own MAP functions and the reduce function to handle massive amounts of data in parallel.

2.3 Application Tier Integration resources

Application Services (SaaS). Software as a service) business, internet-based Business operation has a long history, but also the technology development is very mature cloud computing business. Can be in the application level of railway information system to achieve resource integration, many of the management application software functions are basically the same, such as the day-to-day management of professional software, office management software, e-mail and so on. Each station section installs the Management Server side software, not only configures more hardware and software resources, but also has the more request to the technical ability. These software can be installed in the railway administration's infrastructure "cloud", the implementation of professional management, station section as long as "enjoy" these services can be. SOA technology can be used. Specifically, WebService technology. SOA is an IT architecture, by encapsulating tasks and software functions into a set of interconnected service modules that can be accessed through the network, the flexible combination of software functions is implemented, which provides a feasible solution for the application of cloud Platform software functions. Application-tier Consolidation resources are also important to reduce investment costs, ease maintenance, and quickly deploy usage.

2.4 Network and load balancing.

The network is the Environment foundation of Cloud computing, cloud computing can make the data that transmits on the net grow rapidly, this has put forward higher request to network bandwidth and reliability. The current situation of railway information network can not meet such requirements, we must optimize network layout, improve backbone network speed and increase network redundancy to avoid single point of failure.

Load balancing technology combines multiple servers and services through virtual 1P addresses, scheduling servers and services according to resources. Load-balancing technology is also important for cloud computing data centers.

3. Concluding remarks

The construction of a data center for Railway cloud computing requires strategic decision-making. We should use the more mature technology, start with the establishment of virtualization infrastructure and application services sharing cloud, integrate resources, improve the reliability of information system, reduce the soaring information system investment and maintenance cost, and better serve the railway transportation. Further research on the establishment of railway cloud computing platform services, training professionals, so that based on cloud computing railway Information system has become an intelligent railway transport organization, passenger and freight marketing and management modernization of the strong support, and become a railway to participate in the competition in the transport market, one of the core competitiveness.