How to get started on cloud migration

in the enterprise, many of the successful cases of cloud migration are first migrated from some simpler applications, then moving more applications and data to the cloud in one step, and it is not possible to migrate all applications at once.

For applications and data to migrate on-cloud, a detailed plan and schedule is necessary, and migration is a complex process that starts with the simplest applications and then considers complex, highly correlated businesses, and some personalized enterprise applications.

650) this.width=650; "border=" 0 "src=" http://s3.51cto.com/wyfs02/M01/89/38/wKioL1gNa9WxK93lAAF6F6HIPVA679.png-wh_ 651x-s_142683036.png "/>

As shown in (image from the network), describes the enterprise steady cloud migration of the general steps, migration is a system engineering, migration too fast will often lead to a sharp rise in costs, delays or even failure.

The process of moving up the cloud, we can subdivide it into five steps. Note that the main scenario here is the enterprise private cloud, and its summary steps are also applicable to the private cloud, which is not suitable for migrating to the public cloud. Let's focus on migrating the five phase steps of the cloud.

1. Standardization and Unification

Enterprise traditional IT business applications are generally built on physical servers and storage devices, when the start of cloud migration, the general use of standardized technology, the former server and storage resources to consolidate. It is important to migrate existing old, cloud-ready businesses, and to develop detailed solutions based on the resource situation of the data center; if it is a new application, allocate the appropriate resources and deploy them directly in the cloud environment. Any business that wants to go to the cloud, the assessment of its implementation difficulty is an important means to cloud or transform the risk and benefit assessment of the application system. The process of cloud analysis of the whole business system needs to be carried out from several aspects including hardware support environment transformation, operating system platform change, platform software binding analysis, IP address dependency elimination, API reconfiguration, Modularization transformation, standardization transformation, external dependency condition, etc. To accurately evaluate the related difficulties and pain points of cloud transformation of business information system, we can fully understand and prepare for the cloud transformation of information system.

Of course, virtualization and architecture design are also part of modernizing the cloud business system. Oberyun first without architecture design, because the business will eventually be cloud, regardless of the length of the process of migration, enterprises often use virtual servers instead of physical servers, using storage resource pool to unify back-end storage. In order to realize the management of heterogeneous storage devices, storage virtualization and distributed transformation are often carried out. Of course, in this step, it is possible to involve business transformation consulting and solution optimization, and you must start using scripts or automated installation tools to reduce workload appropriately.

2, procurement or self-built and deployment of cloud services

Virtualization is the first step in Oberyun, and the next step in the migration is to deploy a private cloud management platform. So is sourcing or self-building and deploying cloud services?

Look at the cost and value of the cloud platform. VMware is a commercial software, its maturity and stability has been tested by a large number of real-world environments, but the use of high cost, reflected in its licensing fees and service costs. The benefits of OpenStack free and open are obvious relative to the expensive price of VMware. With VMware's high-input capabilities, OpenStack is available to customers for free. Whether OpenStack or VMware is more valuable is not a clear answer to the question, and the answer depends on the size of the actual deployment of the enterprise. Although OpenStack is free to use, it requires a professional developer and expert in this field, and requires a lot of architectural and building work because it supports many deployment scenarios and the installation process is different. VMware needs to spend some money on licensing and services, and is relatively easier to install and run, and VMware's learning costs are lower, and operations can be easy to get started with.

To be sure, based on the above analysis, large enterprise procurement using the VMware platform is more stable and reliable. OpenStack has a higher entry threshold, and if the enterprise does not have sufficient technical capacity, it will not be able to solve the problems and pits that have been encountered with large-scale deployment of OpenStack.

Building a private cloud requires detailed planning and design as well as implementation, often in the face of resource consolidation and integration and integration of management concepts. In this step, you can also purchase or use public cloud services such as one or more SaaS applications, development testing services, cloud storage, and so on. Hybrid cloud, which combines public and private clouds, is the main mode and direction of cloud computing in recent years. We know that the private cloud is primarily for enterprise users, and for security reasons, companies prefer to store data in a private cloud, but at the same time want to have a public cloud computing resources on-demand expansion, in this case hybrid cloud is increasingly adopted, it will be the public cloud and private cloud mix and match to achieve the best results, This personalized solution achieves both cost-saving and safe purposes.

3. Application migration and data migration

Once the cloud infrastructure and services are deployed, you need to start to unify or upgrade your existing business application services. As mentioned earlier, this step allows you to migrate some of the simpler applications to the cloud and then gradually solve the remaining complex applications.

Application migration process is not a simple point a few buttons to be done, we need to start from the environmental characteristics of the cloud platform, the product of its own adaptation to adjust. For example, whether to support silent installation, disk space usage, parameter settings should be done by the API or CLI, trace and log information through script commands or platform unified collection, etc.

Data migration is the most important for a business application and is directly related to the success or failure of the cloud on the business. Data migration moves files that are seldom used or unused in a business system to a secondary storage system (such as a tape or optical disk), while migrating the data that is commonly used by hotspots to premium storage, such as an SSD or flash array, is a bit like tiered storage management. In order to ensure the security and integrity of the data, the migration of our business is typically combined with a backup strategy and a focused backup of important data. There are business systems on the Cloud to O, replace Oracle with MySQL, then will be related to SQL syntax adaptation, data conversion, new and old system interaction, application transformation and even reconstruction, and so on, these need to be fully considered during the migration phase.

The implementation of data migration can be divided into 3 phases: preparation before data migration, implementation of data migration, and test verification after data migration. Due to the characteristics of data migration, a lot of work needs to be done in the preparation phase, and adequate and thoughtful preparation is the main foundation for data migration. Specifically, a detailed description of the data source to be migrated (including how the data is stored, the amount of data, the time span of the data), the data dictionary of the old system database, the quality analysis of the historical data of the old system, the analysis of the differences between the old and new system data; Establish the mapping relationship between the database tables of the new and old systems, the processing methods of the fields that cannot be mapped, develop and subordinate ETL tools, write test plans and calibration procedures for data conversion, and develop emergency measures for data conversion. The implementation of data migration is the most important step in the 3 stages of data migration. It requires a detailed implementation process flow for data conversion, preparation of a data migration environment, business readiness, completion of unfinished business transactions, or the finalization of the technology involved in data migration, and the implementation of data migration.

The test verification after data migration is the check of migration work, and the result of data test verification is an important basis to judge whether a business system can be formally activated. Through the quality inspection tool or write a check program for data validation, through the commissioning of the new system function module, especially the query, report function, check the accuracy of the data.

Of course, in order to ensure the quality and efficiency of data migration, but also inseparable from the good migration tools. Commercial and open source products each have different characteristics, the choice should be based on the specific circumstances of the analysis. Throughout the current domestic some large-scale projects, in the data migration is more than the use of relatively mature ETL products, in fact, you can see some common points of these projects, including: the migration has a large number of historical data, the allowable downtime is very short, facing a large number of customers or users, the existence of third-party system access, The impact of failure will be wide.

Currently, many database vendors also provide data extraction tools such as Infomover for Informix, DTS for Microsoft SQL Server, and Oracle Warehouse Builder for 0raele. These tools solve data extraction and transformation within a certain range, but these tools do not automatically complete the extraction of data, and users need to use these tools to write appropriate conversion programs to improve efficiency.

Then there is the complex application of the enterprise because of the high degree of business coupling, the dependence on the traditional architecture, generally need a lot of transformation development, such as you want to replace the specific middleware and database and business package software, it may take several years to complete this step. Due to the long time period and the too many uncontrolled risks, it is prudent to conduct a detailed migration assessment of the existing system from return on investment and feasibility.

4. Fully automated

In an enterprise, when a large number of business applications are migrated to the cloud, it is important to use the cloud management platform to automate configuration, approval, service delivery, upgrades, and monitoring of business systems. Continuous automation of existing IT processes is critical, and we want to automate every business on-the-go, from online resource bookings for virtual machines and applications to their delivery, which can significantly reduce deployment time, reduce labor costs, and improve system configuration accuracy and consistency. Although basic automation has begun in the standardization phase, the full automation phase involves the introduction of a large number of scripts, application installers, and automation tools into a process orchestration system where the cloud management platform can be used for service and workflow design.

5. Safety, redundancy, and sustainability of operation and maintenance

Traditional business cloud generally need to pass through a number of aspects of resource supply, delivery services, operations and security processes, such as approval, because many of these processes need to be transformed before cloud services are complete and on-line, automated delivery requires IT security personnel for virtual machine templates, software networks, storage resources, operating systems, The application platform, etc. prior authorization or approval. This phase also takes into account redundancy and scalability, including the ability of servers, virtual machines, applications, and cloud management platforms to continue to function in the event of partial or complete failure of the data center. Security operations and IT governance must also be fully established at this stage, and eventually the five-step cloud migration plan will bring the company to a fully cloud operations state.

Business Cloud is a complex system project, whether it is the old application or the construction of new applications, the moving cloud team need to carefully consider whether the cost and operation of the platform model Matching. At this stage, applying a staged migration may be the only option. Some companies have been successful in using this phased approach to change their traditional applications and to minimize risk while benefiting from cloud computing, which may be the topic of cloud for some time to come.

This article from "Hollows Jie Sun" blog, reproduced please contact the author!

How to get started on cloud migration