Cloud computing and cloud data are advertised as a compelling alternative to internal applications, servers, and storage networks. However, does the enterprise really take advantage of the out-of-the-box "infrastructure as a Service" (Infrastructure-as-a-service,iaas) and software-as-service solutions instead of internal servers and data centers?
A recent report by Forrester, a market research firm, entitled "Enterprise users have not yet prepared for cloud storage," questioned companies ' willingness to deliver their key data to the cloud.
The report notes that while "IT pros are willing to take advantage of the low cost per gigabyte provided by cloud vendors ... However, according to Forrester's 2009-year hardware study, so far this is limited to talking on paper. In the short term, respondents in all regions and businesses of all sizes are indifferent to migrating their data to the cloud. Storage as a service has some long-term development potential, but Forrester believes that ensuring service levels, security, regulatory chains, rental leases and long-term pricing can be a huge hurdle and must be addressed before any meaningful progress is made. ”
The report's author, Forrester analyst Andrew Reichman, is not the only one who thinks cloud development is unpredictable.
Long-term application prospects for enterprise Cloud Dim
Taneja Group senior analyst Jeff Boles said: "At this time very few companies will rely on public cloud services as part of their long-term development path." The cloud may have been started, advanced, and deeply developed, but in the longer term the enterprise must migrate it back to the enterprise more often, so that it can better monitor latency, connectivity, disaster recovery, and security. ”
Boles also points out that cloud computing is different from saving data in a public cloud-the latter can be illustrated with an example. "Usually when the calculation does not need to read, the data stored in the cloud is appropriate and feasible." Therefore, the cloud is an appropriate and tightly protected resilient repository for archived data that is infrequently accessed. ”
When deciding whether to save data in the station or to save data in the cloud, Boles suggests that it decision makers must consider the following issues between decisions:
• Total cost in the data retention period
• The difference between an internal solution and an external solution is in cost and access patterns. "For example, if an external cloud (such as Nirvanix SDN) that replicates data between 5 data centers is closer to the end-user than the private cloud data center, then the internal solution lags behind the external solution," he said.
• Your confidence in the openness of external cloud services and the direction of the road map. "Are you sure that the data is well extracted and sent to other suppliers?" said Boles. Are you sure that the supplier will provide you with this service or product as a core service and that the roadmap will not be offset by other requirements? For example, if a manufacturer delivers the product to a user in the form of a product and a service, will the service user or product user have a greater impact on the roadmap? ”
Boles also advises users to investigate stealth costs and overall management before making a decision.
Boles's long-term use of the cloud is not promising. "The advantage of cloud services is that it offers an unprecedented choice, with enough interoperability in the infrastructure to allow users to mix public and private clouds," he said. With the right private infrastructure (such as VMware), the task of selecting feature features can be delivered to the public cloud service. ”
Understand your access requirements
In the view of storage architect Chris Evans, the main problem with the public cloud is "how is the data saved?" How do corporate organizations retrieve these data? ”
on how to determine if using a public cloud is an ideal choice, Evans says it depends on three main factors: the degree of data security required, its value, and the speed of access you need. "If the data is mission critical and the loss of the data is disruptive to the business, no enterprise will entrust all their data to an IT infrastructure that is not owned by them," he said. ”
Evans also believes that when deciding which data to migrate into the cloud, more should be applied rather than platform-oriented. "E-mail is good for using a public cloud. It does not require high performance, data access is minimal, and in most organizations, losing e-mail is not a big problem. ”
The same criteria apply to other applications to determine whether the cloud is suitable for use. "For example, when a batch report is made, the data is parsed in the cloud and eventually returned to the organization on demand," he said. ”
Companies away from the cloud
In the view of Enterprise Strategy Group senior analyst Mark Peters, the public cloud has the most potential for small businesses and start-ups because "these companies want to avoid the initial cost of building IT infrastructure." He also predicted that the cloud would be used to run major short-term computing tasks.
Neither Peters nor ESG believes that large enterprises will migrate to the public cloud. "Because of concerns about security, privacy and compliance, large companies tend to use the internal cloud as a resource pool that can be dynamically allocated, scalable, and manageable," he said. ”
Peters also lists a number of factors to consider when deciding where to save data:
I/O number: Cloud storage providers are typically charged based on service data traffic (I/O). Some vendors even charge for each API request, such as put, get, or copy. This is another reason why cloud storage is not suitable for intensive transaction processing because the cost of service is probably going to grow in line with this pricing model. This pattern allows users to understand the amount of I/O to any outbound data, prompting them to use technologies that reduce the number of data transmitted.
Regulatory requirements: If your data needs to be tightly controlled (especially the people who access the data), it's best to keep the data internally. The audit model for different providers is different, but even the best vendors cannot fully meet the requirements of the risk management team.
Performance and latency: storing data in a cloud storage site away from the server can increase the latency of the environment. This is also one of the main reasons why cloud storage is not suitable for I/o-intensive environments that require response time control under seconds. But this only accounts for less than 20% of the typical data centers. The other 80% are acceptable to some degree of delay, but the distance from the service provider and the total amount of Internet traffic still affect performance.
Cloud is just part of it planning
Greg Schulz, founder and senior analyst for the Server and Storageio group, also believes that public clouds will be used more by start-ups and companies that require dynamic computing power. "For some people, the use of cloud computing is the same as renting other capital," he said. Its procurement costs may be lower; however, cash flow and other benefits also make the cloud more attractive to small businesses or those that are fast-growing or cash-strapped. ”
Like other analysts, Schulz that the cloud is not omnipotent. "The cloud, whether public or private, is more viewed as a supplementary resource for existing IT resources," he said. I also believe that there will be more and more small businesses using cloud for backup and archive. The cloud seems to present such a trend: small businesses, large business use, and conversely, the larger the size of the institution, the higher the risk. ”
Schulz also believes that cloud solutions "can do a lot" if used properly. But problems, challenges and disappointments will ensue if used improperly, or if companies have unrealistic expectations.
Schulz's advice is: "First and foremost, if you're going to use the cloud to save data as part of data protection, you can view the cloud (public and private) as another tier of IT resources with different performance, availability, capacity, security, and economic characteristics. ”