9 Requirements for storage in cloud computing environments

Introduction

In all recent attention to cloud computing, storage is more regarded as a basic platform. Today, many cloud computing products are limited to the collection of CPU cores, quantitative memory allocation, low-speed storage, or some Internet-Oriented IP technologies. Recently, there have been interesting advanced technologies related to cloud computing and storage, especially the web services access method, making access to storage no longer limited by device files or NFS loading points.

The typical "enterprise-level features" of data storage and management are constantly emerging in IT architecture innovation. Storage architects are aware that these features are important to key businesses and production applications, but currently cloud computing lacks these features. This White Paper aims to describe nine essential elements for storage in enterprise cloud computing.

Factor 1: Performance

Performance costs. In a well-structured application, the performance and cost are balanced. The key to achieving this is to use appropriate technologies to match the performance of enterprise business applications. First, we need to convert the business language of an enterprise into an IT model. Because of this difficulty, enterprises usually stop at static IT architecture and cannot cope with the changing performance requirements of the business. Enterprise cloud computing provides a platform that can better meet the changing performance requirements.

In early cloud computing platforms, storage I/O generally experienced high latency. This is because the vendor noticed that the cloud computing data is easier to access, but did not notice that the service level related to performance, bandwidth, and iops is improved. There are two causes of high latency: Access Mode and type, and Storage Distribution configuration.

The access mode includes a combination of multi-layer protocols (such as soap, NFS, TCP, IP, and FCP) located on the OSI model physical layer ). Data access includes the shared physical service layer (such as Ethernet) and several protocol layers (such as soap or NFS), which generally produces more latency than the dedicated physical layer (such as FC. Most cloud computing platforms on the market include Internet data access, resulting in more data access delays.

For storage media, most cloud computing markets use SATA disks in raid or jbod configurations. Because the performance of SATA (which is regarded as a nearline disk for a period of time) is generally slightly worse than that of an enterprise disk (usually an FC disk), the performance of storage devices is lower than that of applications.

When you use a low-performance storage medium with relatively low bandwidth and high latency access mode, enterprises that use the entire storage subsystem cannot support more key business applications. As a result, this solution is generally only applicable to testing and development.

In contrast, enterprise cloud computing platforms need to provide more options for different performance storage layers. When performance requirements change, for example, from testing to production, the storage platform should be able to use this change. The ideal enterprise cloud computing storage should have a variety of performance zones that can be adjusted to provide the appropriate I/O performance level to the business performance requirements.

Finally, to meet the performance requirements of enterprise-level high-end storage, cloud computing solutions must adopt enterprise-level technologies that are higher than or are currently in use. FC San is generally used. In addition, how to use technology is equally important to the technology itself. In a system management environment, enterprise-level Virtual Machine configurations must continuously provide high performance.

Factor 2: Security

Security and virtualization are often seen as conflicting. After all, virtualization frees applications from physical hardware and network boundaries. Security, in other words, is to establish a boundary. Enterprises need to consider the initial architecture of virtualization design.

In most cloud computing markets, whether public or private, data security is based on trust, which is usually in the management program. When most virtual machines share physical Luns, CPUs, and memory, the management program ensures whether the data is destroyed or accessed by the wrong virtual machine. This is the same as the basic challenges faced by cluster servers over the years. Any physical server that may need to take over the process must have the permission to access data, applications, and operating systems. For example, for external backup, The Lun may need to be mapped to a public backup server.

In enterprise cloud computing, there are two ways to protect business data. First, it involves System Program Management Security. The main goal is to minimize the number of systems used to prevent any virtual machine from negatively affecting other virtual machines. Enterprises also need to protect their Luns from being accessed by other servers, just like external backup servers.

Other important points are data channels. Enterprises must note that they can only provide access paths to physical servers that need to maintain necessary functions. This can be done through npiv (San n port ID virtualization) using zoning, LUN masking, access list, permission configuration.

Factor 3: Automatic ILM Storage

Information lifecycle management (ILM) has become the focus of effective market behavior and is favored by vendors selling Hierarchical Storage. Although ILM is simple in nature-the cost of storage matches the business value of data, the actual challenge comes from the real implementation effect. The granularity of many so-called ILM solutions is not small enough, it is difficult to achieve this goal.

Today, traditional ILM is not deployed on most cloud computing platforms on the market. There are two reasons: first, in many cloud computing, most of the disk media uses the lowest layer storage in a typical ILM solution, so that data cannot be migrated to a lower layer, and ILM cannot be deployed. Second, many enterprises do not need to manage data for a longer time for commonly used computing types of cloud computing, because traditional cloud computing is usually used for function testing/development, concept verification (POC), web server testing, and so on. Taking into account a variety of factors, the fine-grained complexity and cost required for implementing ILM policies do not match the economically saved cloud computing.

According to some industry reports, 70% of the data is static. Enterprises can cut costs by storing suitable data on suitable media. They are aware that deploying cloud platforms can save costs and the economic benefits of implementing Ilm in cloud computing are significant. However, the premise is that you do not need to interrupt the application and do not increase unnecessary operation complexity.

To do this, enterprises must use policy-based block-level ILM methods, regardless of the access method and application type. By tracking data attributes at the data block level, you do not need to perform data archiving or data migration at the operating system level. This method can also be independent from the operating system type and independent from the access method used to store data. It not only optimizes storage costs while maintaining performance (all data is written at the high-speed Layer), but also reduces energy consumption by accumulating unused data blocks to the low-speed layer. This is reasonable because nearline storage only consumes about 20% of the energy consumed by enterprise storage. Therefore, to achieve automatic Hierarchical Storage that is truly suitable for enterprise-level applications, volume-level or file-level data migration cannot be competent, and granularity must be refined to the data block level. Only data migration at the block level can be independent from the operating system type and independent from the Access Method for storing data, so that the application can be supported on demand.

Element 4: Storage Access Mode

There are three mainstream methods to access a bucket: Based on data blocks (FC San or iSCSI), based on files (CIFS/NFS), or through Web Services. Block-based and file-based access methods are the most common in enterprise applications and can better control performance, availability, and security. In this regard, most cloud computing platforms on the market use web service interfaces, such as soap and rest (representative State transmission) to access data. Although this is the most flexible method, it has an impact on performance. Ideally, enterprise cloud provides all three storage access methods to support different application architectures.

Factor 5: Availability

The IT architecture maintenance window is greatly reduced because enterprises need to support users in different time zones and ensure 24x7 availability. Although service level agreements (SLAs) are generally inseparable from availability, it is hard to measure from a business perspective because there are overlapping composite SLAs of Multiple Architectures.

As mentioned above, I/O performance is the first consideration of most cloud computing platforms in the market. If the cloud platform depends on a part of the architecture that is not managed by the internal IT group, the Redundant Architecture part and approach are the best way to reduce the risk of downtime. Although cloud storage service providers continue to increase availability while considering costs, service level agreements on the market cannot meet the needs of key enterprise applications.

In high-end enterprise cloud computing, the storage system can provide enterprise-level storage solutions, including multi-path, controller, different optical fiber networks, RAID technology, and end-to-end architecture control/monitoring, and mature change management process. In the low-end enterprise cloud computing, the storage availability is equivalent to the service level of the cloud computing platform on the market today. To provide the services required by enterprises, enterprise-level cloud storage providers must leverage sound architecture design and proven innovative technologies.

Element 6: Primary Data Protection

Primary data refers to data that runs online. Primary data can be protected by a single technology or multiple technologies. Some common methods include RAID protection, multiple copies, remote replication, snapshots, and continuous data protection.

In most cloud computing platforms on the market, primary data protection is often left to users. Today, it is rare to find that the above methods are used in the mass cloud computing platform because of the complexity and cost of the technology. Some mass cloud storage solutions protect primary data by maintaining multiple copies of data. The entire system runs in a storage without raid protection to reduce costs.

Enterprise-level Cloud primary data protection should be based on internal enterprise-level solutions. When the business impact analysis (BIA) of the solution is required, reliable technologies such as snapshots and disaster tolerance should be ready.

The main difference between the internal enterprise solution and the enterprise cloud storage is how primary data protection is bound in the solution. In order to continue the experience of deploying the cloud environment as needed, various options must be packaged so that the service can be automatically deployed. As a result, a series of bound options can be packaged to meet a large number of requirements. There may be no technology that can match customer needs, such as snapshots and remote replication. In any case, most users will realize that they often need to sacrifice flexibility to gain other management benefits in the enterprise cloud.

Element 7: Secondary Data Protection

Secondary data is backed up by historical copies of primary data. This data protection means to reduce data corruption and restore deleted or overwritten data, so that data can be stored for a long time for business or regulatory purposes. Typical solutions include backup software and storage media. Repeated data deletion may be available, but this may cause some problems in a multi-user environment related to data isolation.

Some commercial and public-sector solutions are added to the mass cloud storage for secondary data protection. However, many vendors of the public cloud computing platform package these together with online storage. Although the reasons vary, in many cases, the service level (SLA) Issues related to the recovery time and retention period are hard to address.

Whether the solution is a private or multi-customer cloud platform, the level of management means, visibility, and recovery services is the key to secondary data protection. Once a recovery request is submitted, the recovery should be started directly and automatically. Users should be able to control the foreseeable recovery performance level (GB/duration of recovered data), and should be able to select a retention period from a short list of options. Finally, you should be able to check the status of those online backups. Because the frequency and retention period determine the resources required for storage and backup-that is, the cost-the customer should be able to observe the resource usage and billing status online to avoid unexpected check.

Factor 8: flexible and adaptable Storage

The flexibility and adaptability of Storage refers to the ability to respond to storage resources as needed based on changes in business needs. Ultimately, it depends on the operating system's ability to check storage changes and the access method used.

Among the nine elements discussed here, this is the best factor for mass cloud computing platforms. Most solutions have a pre-designed quantity to gradually increase storage capabilities. Deleting a space is also an option, which is usually used for volumes or loading points. As mentioned above, the ability of the operating system to respond to storage changes is usually limited.

Storage that supports enterprise cloud needs to be flexibly scalable and billed in a way that the customer understands. Although it is very important to add or delete buckets, users tend to pay only for the space they use. They also want to adjust and generate usage reports in the Web-based management mode. This feature helps them control costs and provide intelligent business planning.

Factor 9: storage reports

When companies consider outsourcing all or part of their IT architecture, they often worry about the invisibility of these specific technologies. The customer needs to understand the running environment status from the perspective of capacity and performance. To achieve this goal, it is necessary to output rich storage reports through the user management interface, so that customers are confident in the efficient operation of storage.

On the cloud computing platform on the market, storage-related reports are the most basic tools. Many providers provide standard reports. In some cases, they also provide basic performance evaluation tools, either from providers, shared vendors, or third-party tools.

Enterprise cloud has advantages over traditional enterprise storage. Traditional Enterprise Storage often leads customers to use a single storage vendor solution. This makes the report very simple, because the data does not need to be translated like from a multi-vendor platform to generate a unified appearance report. Detailed information about the history and real-time usage, together with some key performance indicators-historical or real-time-should be available in real-time 7x24 on the user management interface. In the end, cloud computing providers should have more comprehensive and accurate report capabilities, especially visibility into the usage of storage systems, to reduce enterprises' worries about the loss of control.

Conclusion

A strong enterprise cloud should not only focus on CPU, memory, disk, and IP Address Allocation. When planning the policies of the enterprise cloud computing platform, it should include the nine factors mentioned in this article. In this way, enterprises can have a more complete cloud computing platform to support business operations.