Cloud computing and cloud data storage have been heavily hyped as attractive alternatives to in-house application, server and storage networking. But are enterprises truly scrapping their in-house servers and data centers in favor of off-site Infrastructure-as-a-Service (IaaS) and Software-as-a-Service (SaaS) solutions?
A recent report from Forrester, titled "Business Users Are Not Ready for Cloud Storage," casts doubt on enterprises' willingness to trust their critical data to the cloud.
The report noted that while "IT professionals are eager to take advantage of the low cost per gigabyte offered by cloud providers ... data from Forrester's 2009 hardware survey shows that this is just talk, so far. Respondents in all geographies and of all company sizes appear to have little interest in moving their data to the cloud any time soon. There is long-term potential for storage-as-a-service, but Forrester sees issues with guaranteed service levels, security, chain of custody, shared tenancy, and long-term pricing as significant barriers that still need to be addressed before it takes off in any meaningful way."
Andrew Reichman, the Forrester analyst who wrote the report, is not the only one who sees a partly-cloudy future.
No Enterprise Clouds in Long-term Forecast"Very few businesses are at a point right now where reliance upon a public cloud service is a part of their long-term roadmap," said Jeff Boles, senior analyst at the Taneja Group. "The cloud may be used to start a service, get it staged and well developed, but longer term, the business will often move it back in house so that it can better engineer and control aspects like latency, connectivity, DR and security."
Boles also pointed out that there is a difference between cloud computing and storing your data on a public cloud and that a case can be made for the latter.
"Often data storage in the cloud is actually ideal and more practical when it is not necessary to access it with compute," he said. "Then the cloud becomes an ideal, highly protected, elastic repository for infrequently accessed archival data."
When determining whether to store your data on-site or off it in a cloud, Boles advised IT decision makers to examine several issues before making a decision:
- The total cost over the retention period of the data
- Differences in the costs and patterns of access between an internal solution and an external solution. "As an example, an internal solution may suffer from more latency than an external solution if an external cloud like Nirvanix's SDN is replicating the data between five different data centers that are closer to end users than your private data center," he said.
- Your confidence in the openness and roadmap directions for external cloud services. "Are you confident that the data can be extracted and ported to other providers if necessary?" Boles asked. "Are you confident that your provider is offering this service or product as a core service, and the roadmap won't be pulled in one direction or another by other demands? For example, if the product is offered by a single vendor as a product and a service, do service customers or product customers have more influence upon the roadmap?"
While Boles wasn't terribly optimistic about the long-term adoption of clouds, he did see a silver lining.
"The great thing about cloud services is that they provide options like never before, and there is finally enough interoperability within the infrastructure that users can blend public with private," he said. "With the right private infrastructure in place [such as VMware], select sets of functionality can be handed off to public cloud services, and by handing off selected sets of functionality, costs might be optimized, and the complexity of turning to the public cloud reduced."
Know Your Data Access RequirementsFor Chris Evans, a storage architect for Brookend Ltd. who blogs as the Storage Architect, the main issue with public clouds "is how data is placed there and retrieved by the organization."
As for how to decide if using a public cloud is the right choice for your business, Evans said it depended on three main factors: how secure the data needs to be; how valuable it is; and how quickly you need to access it. "No organization is going to commit their data entirely to non-owned IT infrastructure if that data is mission-critical and if loss of that data would put the company out of business," he said.
Evans also believes that decisions on what data to move onto a cloud will be application- rather than platform-based. "Email is a perfect solution for the public cloud," he said. "It isn't high performance; data access in/out is minimal; and in most organizations, loss of access to email isn't critical."
Those same criteria should also be applied to other applications to determine if they are cloud worthy. "For example, where batch reporting is performed, data can be 'crunched' in the cloud and the results brought back in-house as required," he said.
Enterprises Stay Away from CloudsFor Enterprise Strategy Group Senior Analyst Mark Peters, public clouds currently hold the most promise for smaller businesses and start-ups "that all share a desire to avoid the initial cost outlay of building an IT infrastructure." He also sees clouds being used to run major short-term compute jobs.
What Peters and ESG are not seeing, though, is larger companies moving to public clouds, "because of concerns about security, privacy and regulatory compliance." Larger companies, he said, are looking at internal clouds for "scalability, manageability and resource pools that can be dynamically allocated."
Peters also outlined a number of factors to consider when deciding where to store data:
- I/O count: Cloud storage providers often charge for how much data is moved into or out of the service by the I/O. Some even charge for every API request (such as PUT, GET or COPY). This is yet another reason cloud storage is not suitable for heavy transaction processing environments, as service costs would soar with such a pricing model. This model also makes it imperative for users to understand the I/O profiles for any data they store offsite, and use technologies that reduce the overall amount of data transferred.
- Regulatory requirements: If data is highly regulated (especially regarding who can access the data), it's almost always best to keep it in house. The ability to audit access varies by provider, but even the best might not completely satisfy risk management teams.
- Performance and latency parameters. Storing data far away from the servers at a cloud storage site introduces latency into the environment. This is one of the primary reasons that cloud storage is not suitable for heavy I/O environments that demand sub-second response times. But that's less than 20 percent of the data in a typical data center. The other 80 percent is somewhat latency tolerant, though distance from the service provider and overall Internet traffic can still affect performance.
Clouds Just Part of Bigger IT PictureGreg Schulz, founder and senior analyst at The Server and StorageIO Group, likewise sees public clouds being used more by startups and enterprises that require dynamic compute capabilities that are looking to defer or shift costs. "For some, the use of cloud compute is similar to leasing other capital items," he said. "It might be cheaper to buy; however, the cash flow and other benefits can [make clouds] more attractive for smaller or rapidly growing or cash-strapped businesses."
Schulz, like other analysts, doesn't see clouds as an all-or-nothing value proposition. "Rather, clouds, both public and private, are increasingly being seen as complimentary tiered resources to existing IT resources," he said. "I'm also seeing more smaller companies leveraging clouds for backup and archiving on a larger scale. The trend seems to be the smaller the organization, the larger the adoption; the larger the organization, while there are isolated and marquee banner adoptions, the more risk averse."
Schulz also argued that cloud solutions "can and do enable good things, when used appropriately." However, when used in inappropriate ways, or when enterprises have unrealistic expectations, issues, challenges, problems and disappointment can be expected.
Schulz's advice: "First and foremost, if you are using clouds to store your data, include them as part of your data protection. And look at clouds, both public and private, as just another tier of IT resources with different performance, availability, capacity, security and economic attributes."
Follow Enterprise Storage Forum on Twitter