Shulou(Shulou.com)06/02 Report--

The full text of this article is about 5000 words and it takes about 7 minutes to read. The following is the text.

Digital transformation is having a far-reaching impact on enterprises. In the face of higher requirements for enterprise data centers in terms of performance, agility and total cost of ownership put forward by business digitization, Internet and intelligence, traditional data centers have become increasingly difficult to deal with. Through the "cloud" to complete the modernization of the data center transformation, so as to support business innovation has become a consensus in the industry.

Superconvergence is becoming the mainstream solution for building the "cornerstone" of a private cloud, but do you know the relationship? In this article, we have introduced the main components of the private cloud solution and explained the importance of the IaaS "cloud" transformation of the enterprise data center. This article will further explore some key elements and implementation schemes for the successful IaaS upgrade of the enterprise data center.

How to define the success of enterprise data center "cloud"?

In the face of many products and solutions in the market, many users want to confirm whether their chosen technology route is correct and reasonable in the process of transforming their enterprise data center "private cloud". In response to this problem, Gartner, a well-known authority, has given a lot of valuable views on global and Chinese private cloud construction in China Summary Translation: 'Rethink Your Internal Private Cloud'. The first conclusion is "Many organizations claim to have adopted an internal private cloud. However, the definition suffers from being widely misused." (many organizations claim to have adopted a private cloud, but this definition is actually being misused), and many of these users interpret the use of OpenStack or the mere completion of a virtualization transformation as a private cloud transformation. For this reason, Gartner gives the following preliminary definition in this paper.

The above is an introduction from Gartner "Rethink Your Internal Private Cloud"

In fact, unlike public cloud users who only need to focus on resource use and payment, private cloud requires not only scalability and flexibility in resource delivery, but also a greater challenge: the overall solution needs to meet the following core requirements from planning, deployment to delivery, and operation and maintenance:

On-demand investment: different levels and different scales of on-demand investment can be made according to the actual needs of the business. For example, the phased construction of the IaaS layer and the PaaS layer, the separate construction of the management platform and the resource pool in the IaaS layer, and the gradual expansion of the resource pool according to the business situation; rapid launch: private cloud not only requires rapid delivery of resources, but also requires automatic deployment of business systems and convenient online expansion; OPS is simple: in the private cloud scenario, the enterprise is both the user and the operator, and the OPS staff is very limited. Therefore, the operation and maintenance of private cloud should be as simple as possible, while ensuring that the complexity of the system does not increase significantly with the expansion of the system scale. This can prevent complex operation and maintenance from becoming a bottleneck for rapid business growth and a hidden danger to the stability of the system; production readiness: when users carry out private cloud transformation, they usually only focus on functional features. while ignoring that the private cloud ultimately has to host production business in order to truly achieve the goal of enterprise data center "cloud" transformation. Therefore, the "production-ready" capability of solutions and even supplier services can ensure that the architectural transformation is sustainable and in depth, rather than just staying at the experimental stage; the cost is reasonable: all transformations ultimately need to be measured in terms of economic benefits. Private cloud solutions need to pay attention not only to procurement costs, but also to use costs, maintenance costs, software and hardware maintenance and upgrade costs.

Only when the transformation of enterprise data center fully meets the above requirements can it be considered that the goal of "cloud" has been initially achieved.

The key points of the Transformation of Enterprise data Center "Cloud"

"if you want to fully meet the above needs through IT infrastructure upgrades, it is not enough to use traditional architectures or simply add cloud management, for example, the following more comprehensive requirements for the cloud transformation of bank data centers:"

Adapt to the needs of elastic change and rapid deployment of computing resources in the Internet environment; carry out cloud computing architecture planning and formulate cloud computing application strategies; explore the construction of private cloud platforms, using high maturity and open computing virtualization, container virtualization, distributed storage, network virtualization and other technologies; establish a resource pool to form a private cloud platform with flexible supply of resources, flexible scheduling and dynamic measurement

……

Combined with relevant industry reports and a large number of user practices, we have also sorted out the following key points for the "cloud" transformation of enterprise data centers:

Transformation point 1: from resource isolated island to real resource pooling

In traditional data centers, servers, storage and other resources are difficult to share and flow due to various reasons (isolation policy between business systems, devices of different brands and models, etc.). The utilization rate of a single resource is easy to be polarized (ultra-high load and no load), but in fact, the overall resource utilization is relatively low.

In fact, resource pooling is to virtualize computing, storage, network and other resources through virtualization technology to form a unified resource pool at the data center level, where resources and data can be easily shared and flowed. at the same time, different degrees of isolation can be made through policies. In this mode, application systems and business departments can obtain resources on demand and conveniently, reduce the waste of resources, and effectively improve the overall utilization rate of resources.

Transformation point 2: transition from centralized to distributed architecture

In the traditional data center, both the application layer and infrastructure level are centralized, and the centralized architecture is extended through scale-up (enhance the hardware configuration of a single device). In this expansion mode, the system performance is easy to reach the bottleneck and the scalability is not good. In addition, there is another problem with the centralized architecture, which depends on the reliability of a single device, the system is not redundant enough, and there is a risk of a single point of failure.

In fact, many enterprises are considering transforming their business and infrastructure to a distributed architecture, which has the advantage of not only high concurrency performance, but also easy scale-out performance and capacity, and good fault tolerance.

Transformation point 3: transition from dedicated hardware to open software-defined model

In the traditional data center, there are a large number of servers, storage, network and other hardware devices, including some special hardware-based devices, such as mainframes, minicomputers, networks, storage devices and so on. These devices are often complex and expensive, and the upgrade is slow. In addition, with the expansion of the scale of the data center, different types of equipment will greatly increase the complexity of the system, which will bring great difficulty to the management.

Through mature software-defined products, software and hardware can be decoupled, the shackles of special hardware can be eliminated, hardware standards can be unified, and management can be simplified; software-defined technology enables users to obtain more and newer functions easily without replacing hardware. The software-defined model also promotes the faster application of more advanced hardware technologies in the data center; software-defined products use general server hardware instead of dedicated hardware, so that the total cost of ownership of the product will be greatly optimized.

Transformation point 4: from manual processing to self-service and automatic service

In the traditional process, there are many steps in application development, testing and launch that need to be handled manually by developers, testers and IT administrators. The whole process is relatively complicated and staggered, and the efficiency from development to release is relatively low.

The transformed business units and developers can apply for virtual resources by themselves through electronic workflow, and the system invokes automated deployment tools and customized templates to automatically complete system configuration and application software installation, as well as related resource changes and other operations, which can effectively improve the efficiency of application development and resource distribution.

Transformation point 5: the transformation from decentralized statistics to unified measurement

Before the completion of resource pooling in the data center, it is difficult to get a unified view of the real-time use of resources in the data center, and IT managers have no way to accurately analyze the input and output of resources and make reasonable budget and planning.

Under the premise of resource pooling, combined with the resource measurement function, it can reflect the resource usage in real time, accurately control the resource quota, and count the team resource usage; and even automatically output capacity forecasts and suggestions according to the historical data of resource use. to help users make reasonable resource expansion plans.

The above transformation can be easily achieved by using super convergence combined with cloud management software.

In the previous article, "Super convergence is becoming the mainstream solution for building the" cornerstone "of a private cloud, but do you know the relationship? In this article, we have introduced the relationship between superconvergence and private clouds. It can be seen that the super-convergence architecture combined with the cloud management platform can well meet the core needs of the "cloud" transformation of the enterprise data center.

Realizing the Transformation of Resource Pool by Super Fusion

1. Pooling of resources in one step

The super-converged infrastructure integrates the resources of multiple servers in a software-defined way, virtualizes computing, storage, network and other hardware resources respectively, and builds a complete and unified infrastructure resource pool. In short, super-converged software can complete the virtualized resource pool construction of the infrastructure at one time.

two。 Fully distributed architecture

The core of the scheme is distributed storage, which supports horizontal expansion, and strong concurrent processing capacity can be obtained by adding nodes to cope with the changing business requirements. The reliability of the super-converged infrastructure does not depend entirely on the hardware itself. the system provides high-level fault-tolerant design, supports high availability and fault self-healing, and can effectively reduce downtime.

3. Software definition

The super-converged infrastructure uses standard hardware (commercial servers and Ethernet switches) instead of dedicated hardware to unbind hardware and software. At the same time, hardware standardization itself can greatly reduce the complexity of management. Software-defined features enable the super-fusion architecture to iterate quickly to obtain more functionality, support the latest hardware, and fully release hardware performance.

4. Converged deployment

In addition to the above transformation points, another important feature of super convergence is the converged deployment of distributed storage and virtualization, which further reduces the complexity of IT architecture and computer room footprint, and reduces the total cost of ownership.

Automation and measurable transformation by cloud management platform

Cloud management platform (Cloud Management Platform, referred to as "CMP") is built on super-integration, which is responsible for linking developers, IT administrators, and unifying the measurement and planning of resources. The main functions related to the IaaS layer include:

1. Automation and self-help Portal

CMP provides a self-service portal, and developers can initiate a self-service process, which can be integrated with the enterprise's OA or ITSM to complete resource application and approval, and automatically complete virtual machine creation and related configuration changes by calling HCI-related APIs. This can avoid a large number of complicated manual operations and greatly improve development efficiency.

two。 Transparent resource management

CMP can control the disorderly distribution of resources through resource quota and resource application process, and provide real-time usage reports and capacity planning suggestions for IT managers. Through virtual machine life cycle management, the resources that are not in use can be recovered in time and the waste of resources can be reduced.

3. Multi-cloud management

Private cloud and public cloud can not replace each other, both have unique value. "hybrid cloud" will be the "standard" of future enterprises. It is necessary for CMP to provide multi-cloud management because it can help enterprises extend applications from private cloud to the edge of public cloud and migrate applications between public cloud and private cloud. This strategy will help enterprises significantly reduce costs in the future.

Practice of SmartX scheme

SmartX super-convergence products not only have the standard advantages of the above super-convergence architecture, but also have the following unique advantages:

Distributed block storage has been independently developed and tested by a large number of financial customers' production environment, which can effectively support users to carry out the overall transformation of enterprise data center. The product has good openness. It not only provides free native virtualization service ELF, but also supports VMware and Citrix virtualization platforms. In terms of server compatibility, it not only supports the mainstream x86 server brands commonly used by users, but also provides domestic server support. SmartX super convergence also has a leading edge in terms of performance and the use of computing resources.

In addition, SmartX also provides a distributed block storage deployment model with separate storage and calculation, which flexibly supports the distributed and software definition transformation of user storage resources.

Based on SmartX product line and CMP, a stable, simple, flexible and open IaaS layer scheme diagram is formed.

At present, SmartX has established ecological cooperation with a number of professional CMP manufacturers in the industry, and completed the docking of cloud management platform to super-fusion platform. The docking solution further forms the following unique advantages:

1. Stability: whether it is super-integration or CMP platform, its core is by domestic professional manufacturers and has been tested by many industry head customers' production environment respectively, which can really support users to realize the overall transformation of enterprise data center.

two。 Openness: the whole solution has good openness. Users can carry out super-convergence and cloud management construction in batches according to their actual business needs. Professional cloud management manufacturers can integrate users' original VMware systems. SmarX super-convergence also supports a variety of virtualization platforms and server hardware.

3. Simplicity: SmartX hyperfusion implements the transformation of resource pooling, distribution, and software definition patterns in the most concise way, while the CMP platform is easier to use than the complex components of OpenStack.

4.。 Investment on demand: both super-convergence and CMP platforms can be built on a very small scale, making it easier for users to start the transformation and truly realize the concept of "Think big,Start Small,Move Fast".

At present, the program has formed an actual case in a large domestic state-owned bank and a large domestic securities institution, and we will further introduce these user practices in the future.

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.