Cloud storage: Opportunities for hidden challenges

The technical maturity makes the practical cloud solution both available and affordable. At present, most large companies are already exploring ways to make enterprise data centers more "like" the cloud, thereby increasing efficiency, cutting costs, and providing the flexibility required to adapt to rapidly changing business needs.

Cloud storage caters to new trends

Whether it's a public cloud or a private cloud, the key to success is building a proper server, storage, and network infrastructure in which all resources are used and shared efficiently. Because all data resides on the same storage system, storage is particularly important in shared infrastructure patterns.

It is already clear that, in terms of computing, server virtualization provides a suitable infrastructure for cloud services, because this technology, when needed to change, can effectively divide computing resources and quickly allocate, increase, reduce, or unassign resources. A series of rapidly maturing virtualization management services can also help improve speed, flexibility, and higher availability. Leading cloud service providers such as Amazon Web Services (AWS) are already using this approach to take full advantage of the latest virtualization technologies.

The pressure to reduce storage-operation costs and to spend less is greater than ever, due to the economic downturn, especially in the face of accelerated data growth. Traditional storage technology is not designed for use in the multi-PB Web 2.0 era, and if a traditional storage architecture is used, a new storage array will be added when capacity requirements increase. As the number of arrays that need to be managed grows, the storage environment becomes increasingly complex, management is more difficult, and higher operating costs are required. This has brought more disadvantages to the business-extending time-to-market, reducing productivity, and weakening flexibility.

A concomitant challenge is that file-based data is growing much faster than block based data, due to the explosive growth of digital content. By 2012, more than 80% of the storage capacity will be used for file data, according to industry analysts. This is not just for primary storage systems, but also for systems that store copies of data for use in data protection, disaster recovery, test development, archiving, and collaboration. While traditional storage technology continues to perform well in areas that are inherently good, that is, transactional computing, this solution is powerless to curb the growing trend of file-based data. These factors are prompting users to consider new storage deployment patterns, such as cloud storage.

Cloud storage is provided as a service through subscription mode. A service provider can be a company's internal IT group (a private cloud), a third-party company that specializes in providing storage services (a public cloud), or a mixed form of both (a mixed cloud). The economic benefits of cloud storage benefit both service providers and enterprise customers. Service providers gain economies of scale through a multi-tenant infrastructure, as well as predictable recurring revenue. The benefit of the enterprise is the ability to dynamically allocate and deallocate storage resources, provide level-appropriate storage capacity and data protection, and thus have the flexibility to extend storage.

Requirements for cloud storage

In general, cloud storage must be resilient, and the underlying infrastructure should be quickly adjusted to meet changing needs. In addition, automation is needed so that policies can be fully leveraged to quickly change the underlying infrastructure without human intervention.

To provide a seamless, easy to manage elasticity mechanism, the cloud storage service solution must meet a number of requirements if the benefits of enterprise customers are to be fully met: cloud storage needs to be rapidly expanded to provide huge storage capacity. This means that regardless of the object (billions of), performance, user, client (thousands of virtual servers that access storage resources in parallel) and capacity (up to petabytes), the flexibility to scale, while a single namespace provides all storage capacity, is important to reduce operating costs.

In a cloud storage and shared data center environment, data privacy has become a priority concern. Customers of cloud storage services often decide to store their data on a partition in a shared storage system, which alleviates the need for cloud storage providers to purchase and manage specialized hardware. While a shared storage infrastructure can reduce costs, customers want to ensure that their data is private and not visible to other customers. Cloud storage providers must develop management policies for multiple tenants so that multiple business units or different companies can securely share the same storage hardware.

Provide a mature and stable storage infrastructure for customers to quickly and reliably recover data, a necessary service for cloud storage providers. Enterprise customers expect their storage data to be available immediately and available 24x7, which requires a cloud storage infrastructure with a long average of no downtime (MTBF) and average data loss time (MTTDL).

Enterprise users also need to ensure that their data is backed up reliably for disaster recovery purposes and that the relevant guidelines are met. Due to the high cost and technical expertise required, IT departments often overlook disaster recovery and guidance, such as payment card Industry data security Standards (PCI), SAS70 Auditing standards and health Insurance Portability and Accountability Act (HIPPA), and so this is one of the most important services that cloud storage providers can offer. Cloud storage providers must automatically replicate customer data to one or more data centers, in accordance with service level agreements (SLAs). Such policies clearly set out how long it would take to ensure data availability in the event of a disaster.

One of the big benefits that many companies now expect from deployed cloud storage scenarios is that they can improve manageability in the face of dramatically increased storage capabilities and costs. If the Amazon Network Service sells simple storage service (S3) cloud storage capacity for 12 cents a month, it's not hard to imagine what the internal costs are, or how much the enterprise deploys private cloud storage or service providers to implement public cloud storage. Fundamentally, the cloud storage provider model can certainly manage more than 1PB of storage capacity and significantly reduce operating costs, which is not possible with an enterprise storage deployment model.

Secure data access

The use of cloud storage by enterprise IT departments also requires flexibility. When it comes to storing enterprise data offsite, the ease with which an external storage service provider does or claims to be is still different for enterprise users. As businesses need to eliminate data security concerns and have some confidence in the cloud storage model, cloud storage has a public, private, or mixed three deployment pattern.

The IT department is responsible for multiple cloud storage platforms, determining which specific types of data fit in the location-some in the enterprise and some outside the enterprise. In a mixed cloud, infrastructure and policies allow multiple clouds to be managed as a whole.

An important aspect of cloud storage is easy access to data in the cloud, which is critical for cloud storage to seamlessly integrate into existing enterprise workflows and minimize the difficulty of using cloud storage. The current focus of cloud storage providers is the ability to store Internet applications on the network by enabling protocols such as Representative State Transfer (ReST). There is a growing interest in this kind of service. The improved, rest-enabled Amazon S3 application programming interface (API) is likely to become a de facto API for accessing Internet services through cloud storage.

However, requiring an enterprise to use a programming interface to access cloud storage can significantly increase management costs because IT managers are not keen on the idea that in order to take full advantage of cloud storage, they need to train their employees in new protocols, or learn how to write programs for proprietary cloud storage APIs.

Enterprise users require standard NFS and CIFS file access protocols to access data in the cloud, so that once data is written to a specific NFS or CIFS file system supported by the storage cloud, users can read, open, and change files, as if they were operating on a local NAS system. This may be possible, for example, by deploying a new type of cloud storage "ramp up" user-side equipment (CPE). Such devices can abstract the APIs and protocols supported by cloud storage providers so that these APIs and protocols are transparent to IT administrators.

In the foreseeable future, there will be a variety of cloud storage scenarios. Cloud storage client devices also allow enterprises to transparently take advantage of the best cloud storage scenarios in their class, while eliminating the risk of being locked by providers. In addition, operations such as validating, encrypting, compressing, and ensuring data integrity can be implemented in a client device to enable access to multiple cloud storage providers in a simplified and secure manner.

Application barriers

Enterprises face a number of risks in adopting cloud storage processes, involving shared leases and data migration/integration issues, contract issues, and IT staff's conservative attitude towards outsourced storage operations.

Cloud storage users agree to put their data together with competitor data, which is why validation and encryption mechanisms must be foolproof and well-defined to mitigate this risk. Quality of service is also a major concern in the Multi-Tenant model, with many companies concerned that a surge in demand that is not relevant to the enterprise may cause performance degradation or service outage.

When data is placed in the cloud, there is also a significant risk that it may be difficult to control the data and that it is more difficult to integrate applications with data. As mentioned earlier, if a cloud storage provider requires an enterprise to use a programming interface for accessing data, this is a major risk to enterprise cloud storage deployments. Companies will be concerned that if they hand over data to control and move data out of the enterprise, this will disrupt application integration projects such as information management and data warehousing, making it a major challenge to update migration data.

Last but not least, storage administrators are reluctant to adopt new technologies or models, because technologies that bring greater risk or uncertainty tend to slow down, and most businesses want to see what happens to well-known peer companies first. It is worth mentioning that the first companies to cloud storage migrations will be companies that cannot run their business through traditional storage enterprise cost structures, such as Web 2.0 companies, which have file data, free product or ad-supported products, and consumer/SME backup mechanisms.

In short, cloud storage is likely to change the pattern of enterprise storage markets in a way that traditional patterns cannot. Data management as a service can significantly increase ROI and meet increasing requirements for controlling the cost of data management. However, major challenges need to be overcome before the new model is applied to mainstream enterprises. The initial development of cloud storage in the "best" deployment environment will change the storage industry over the next few years.

The best areas of application for cloud storage

Not all enterprise applications are suitable for using cloud storage. In fact, for performance-oriented applications, the need for application servers and storage systems in the same place is not changed by cloud storage. Although Internet bandwidth has increased, the increase is not large enough, such as the inability to allow transaction database applications outside the storage system to run. This means that cloud storage is either part of a larger entire application that is provided through the cloud, or a focus on a separate storage system.

For cloud storage, the most mature deployment object is a well-defined application, so it is easy to deploy it as an organic whole (such as using software, servers, storage, and networking) using cloud-based, managed software as a service (SaaS) model or platform-service (PaaS) model. internet-based e-mail or customer relationship management (such as Salesforce.com) is an example of the entire application, which is very limited in its interaction with non-cloud applications and is therefore easily available through the cloud.

Making full use of the whole application based on cloud brings remarkable benefit to the investment return of the enterprise. First, the enterprise is able to offload the entire application series (including software and hardware) to the cloud application provider, and secondly, adding a new application user's incremental cost and time is actually zero, which can greatly improve the efficiency and productivity.

Storage-intensive business continuity and disaster recovery applications are also ideal for using cloud storage. Typically, best practices in business continuity/disaster recovery require backup and archive content to be replicated offsite, many companies require advanced business continuity/disaster recovery capabilities, such as meeting industry oversight, risk mitigation, and business partners, without having to develop the actual venues or expertise they need. As a backup/archive of cloud services, medical image archiving, digital content delivery, video surveillance and data warehousing have evolved from managed-type services to shared tenant models, while specially tailored high-level architectures provide archiving, retrieval and search and long-term retention capabilities to support access to video and large objects. A customer system that leverages cloud-based business continuity/disaster recovery services can be either a single server or a server farm that provides access/playback, optimization, and management capabilities.

The third major area for using cloud storage is to store applications through the web, such as the home directory, CAD/CAM, earthquakes and manufacturing. Most file storage relationships with application servers are loosely coupled or completely decoupled, and users typically connect to the file storage system over the network and then browse and access files. Unlike transaction applications such as online transaction processing and e-mail, which require high-speed access to block storage, the speed at which file-based storage is accessed is not a critical requirement. In addition, businesses with many remote offices are already using wide-area, WAN-based file storage, which makes it extremely appealing to use cloud-based file storage.

Cloud-based business continuity/disaster recovery and file storage services can also enable enterprises to leverage the storage multi-tenant architecture, to address the rapid growth of unstructured content to the cloud storage services to deal with, the investment return to achieve significant benefits.