Cloud storage has brought a big change in the storage structure

Cloud computing is becoming more and more popular. Will this subvert the existing network storage architecture? Now no one can say with certainty what the trend is. However, one thing is certain: Cloud computing has posed an unprecedented challenge to storage capacity, storage access performance and cost.

The emergence of cloud computing has promoted the development of clustered storage technology

What is the impact of cloud computing on information storage? Ren Yuxiang, chief architect of EMC's China R & D center, said: "The starting point for cloud storage should be PB."

Rise of cluster storage

IDC research shows that from 2006 to 2010, the total amount of global information growth increased more than six times, from 161 EB to 988 EB (1 EB = 1024 PB). Some of the newer disk arrays have generally adopted 750GB or 1TB SATA drives. Currently known as the highest storage density disk array can provide up to 42TB of storage capacity in 4U space, which in the past is simply unimaginable.

The latest generation LTO-4 tapes also have a single-cartridge tape storage capacity of 1.6TB (2: 1 compression ratio). The continuous improvement of technology will surely promote the development of storage to higher capacity. The introduction of technologies such as deduplication and compression will further improve the utilization of storage space. From the performance point of view, FC disk array has been gradually transition to the 4Gb era, and 8Gb FC wavers to the data center users; Gigabit Ethernet storage is no longer on paper; in the field of InfiniBand, there are already manufacturers introduced 40Gb InfiniBand adapter products.

Existing network storage architecture, such as SAN or NAS can also effectively support ubiquitous cloud computing environment? Some people expressed doubts. The main argument is: in the face of mass storage requirements of the PB level, the traditional SAN or NAS bottlenecks in the expansion of capacity and performance; cloud computing this new service model will inevitably require the storage architecture to maintain an extremely low cost, and now Some high-end storage devices obviously can not meet this demand. From Google's practice, they did not adopt the SAN architecture in their existing cloud computing environment. Instead, they used the Google File System (GFS), a scalable distributed file system. This is an efficient clustered storage technology.

The gradual rise of clustered storage technologies in recent years has not only easily broken the performance bottleneck of the SAN but also enabled the linear expansion of performance and capacity, which is a new choice for enterprise users pursuing high performance and high availability. With the rapid growth of some companies specializing in cluster storage business such as Panasas, Isilon and Longcuo in the Chinese market, the application of cluster storage technology will become more widespread. Although the advantages of clustered storage in dealing with unstructured data are obvious, cluster storage is unlikely to completely replace the traditional network storage in a short period of time, and SANs and NASs still have their applications.

It is important to emphasize that virtualization is a core technology in achieving the cloud computing vision as cloud computing itself is a next-generation computing platform that offers virtualization and high availability. From the current market conditions, server virtualization is already in full swing, while the development of storage virtualization is relatively slow. The end of 2007, EMC introduced the SAN storage virtualization product Invista 2.0. Compared with the previous generation, Invista 2.0 supports 5x more storage capacity, further improving usability, enhancing data protection and management functions, improving usage efficiency and enhancing scalability. In addition, Invista 2.0 also passed VMware certification, allowing users to better manage, share and secure information within VMware's infrastructure.

Craig Nunes, vice president of marketing for storage at Storage 3PAR, said: "In order to effectively support cloud computing, there are several key features that an infrastructure must have: First, these systems must be autonomous, that is, they must be embedded with automation technologies that eliminate manual deployment and Management, allowing the system to respond intelligently to the application's own requirements.If the system requires human intervention to allocate and manage resources, then it can not fully meet the requirements of cloud computing.Secondly, the cloud computing architecture must be agile, and can be on the demand signal or In other words, embedded virtualization and clustering technologies must be able to cope with rapid changes in business growth or service level requirements.If the system takes hours, days, or weeks Time to respond to new applications or user needs, then the system will not be able to meet the cloud computing requirements. "

SaaS is also an abbreviation of Storage as a Service, meaning storage as a service. In the cloud computing environment, storage is no longer a cold hardware device, but a service. This will not change the way the user's future storage procurement, procurement hardware from the purchase of storage services? Craig Nunes said: "In a large enterprise, end-user IT requirements will ultimately be served by service, whether it's adopting a cloud computing model or building a common data center."

EMC is optimistic about cloud storage

Some time ago a rumor online that EMC and SAP will work together to improve EMC's SaaS strategy and the use of a called EMC Cloud hosting services. As this is not EMC's official announcement, Ren Yuxiang declined to comment. However, EMC into the field of cloud computing is indeed a fact.

In February 2008, EMC announced the acquisition of cloud computing vendor Pi. This is a company that specializes in software and services related to personal information management. Pi's products and services include a series of cloud computing features, such as distributed user data replication. After completing a series of recent acquisitions, EMC has reorganized its internal organization by increasing its division from six (including storage, content management, VMware and RSA) to six, with two new divisions of consumers / Small Business Products and Cloud Infrastructure and Services. According to Ren Yuxiang introduction, currently only EMC China R & D center there are more than 100 technical staff specializing in cloud computing related research and development work.

Online storage service is a typical application of cloud computing. According to IDC's forecast, the global market for online storage services will reach $ 715 million in 2011. Also a company acquired by EMC, Mozy, has started offering online backup services two years ago and now has hundreds of thousands of paying customers, including not just individual users, but also large enterprise customers like General Electric. Fan Chengong, EMC vice president and general manager of R & D Center in China, is a loyal Mozy user.

Ren Yuxiang stressed: "The key to the popularity of cloud computing lies in the user experience. For end-users, they do not care about what technology is actually used in the background, and the result they want through simple operations is what they really are One day, no matter where you go, just carry a little USB-like little thing with personal security authentication information, just plug it into any terminal device and you can move from the cloud Get the information you want. "

Faced with standardization issues

When I first came into contact with the concept of cloud computing, many people naturally think of grid computing and utility computing concepts. It seems that these technologies have more or less some cloud computing shadow. However, some experts believe that cloud computing is different from utility computing and grid computing for the following reasons: Grid computing usually refers to a resource pool environment for computing tasks (such as image processing); utility computing refers to a To master the resource pool environment for long-running processes and tend to focus on meeting service levels; cloud computing in most cases refers to the different types of services provided on the network, and cloud computing can run on grid computing or utility Computing environment.

"From grid computing, utility computing to cloud computing, take the spiral of a rising road of development.C cloud computing clearly put forward and emphasized the importance of virtualization technology, which in the past is not." Ren Yuxiang explained "The concept of cloud computing is truly impressive: the so-called cloud is in the air, meaning that the relevant computing or storage work is not done on the local device, but in the air or remotely."

Understanding cloud computing can capture two key elements: first, cloud computing uses distributed computing; second, cloud computing provides a new business model that has advantages in people, money, services, etc., And large-scale third-party service providers will be able to achieve large-scale operation. Ren Yuxiang said: "The essential feature of traditional computing is that both computing and storage are done locally, and enterprises that adopt traditional computing methods will permanently retain the relevant software and hardware, while the cloud computing method is not provided by third-party service providers Computing and storage resources, and is responsible for operation and maintenance, users can get the services they need by simply plugging in the system with some standard end-use tool, such as a browser. "

At present, the research on cloud computing in the industry is still in its infancy, and the rapid development of cloud computing depends on the development of standardization to a certain extent. Otherwise, communication barriers between different vendors' "clouds" and "clouds" may occur. Although the application of cloud computing lacks comprehensive technical standards, we also see some vendors have already made helpful attempts in this regard. For example, 3PAR unveiled a development plan called 3cV that combines 3PAR InServ Storage Servers, HP C-class Blade Systems and VMware Architectures to create a complete common computing product that can help customers move servers, storage and With half the operating cost reduction and better sensitivity, Deutsche Bank has successfully implemented the program.