|
 |
|
General Information
|
Add To My Personal Library |
April 10, 2009
• Vol.31 Issue 12
Page(s) 28 in print issue
|
Researching Cloud Computing Offerings
Embracing The Cloud Is Not Easy, Especially When The Idea Itself Is Evolving
|
In a data center, moving targets are always the hardest to track. With cloud computing, not only is the target “moving” and evolving, but it’s also often hard to define. For some, it means storing all applications and data in a remote data center. Others define it as offloading a few noncritical services to a provider who can assure data integrity. The cloud often means abstracting data from the hardware layer, akin to virtualization, but it could also be the future of all data processing.
To help you develop a sound cloud computing strategy, we tapped several experts to share their insight and knowledge, with guidance meant specifically for a small to medium-sized enterprise considering the cloud.
 Offload Extraneous Services
No matter how you define the cloud, most experts view it today as a way to off-load services to a provider, as opposed to a technology that warrants wholesale replacement of servers or storage—even in a small enterprise. There are issues related to security and privacy, and the field of cloud computing is still too new as an industry for a company to consider replacing infrastructure with the cloud.
“SMEs should look at receiving services from the cloud that might otherwise be too complex for them to set up themselves or become much more affordable through greater economies of scale,” says Kristof Kloeckner, a CTO at IBM in charge of cloud computing. “Good examples are business resiliency and information protection services, as well as collaboration services. Utilizing overflow capacity in the cloud for highly variable or seasonal workloads is also an attractive opportunity.”
Security Is An IT Concern
According to Simon Crosby, the CTO at Citrix (www.citrix.com), companies that choose to use cloud services should realize that security is an internal concern, not an external issue. Cloud vendors may provide assurances about data security, but the ultimate responsibility falls on the IT department and data center, not on the vendor. For a small to medium-sized enterprise, it might be easy to hand off security management as part of the cloud service. For example, if a company decides to use a cloud provider for a parts inventory system, the SME should be highly involved in terms of authentication, passwords, encryption, and backup for that inventory system and not just rely on the provider to make sure the database is secure. In a discovery for litigation, saying that the company thought the cloud provider would secure the data would not be a good defense—it’s important to show how internal IT is involved.
“It is the responsibility of the organization, not the service provider, to secure valuable data,” says Crosby. “Likewise, detailed service-level agreements need to be put in place to ensure reliability and sustainability for a technology that is outside of the enterprise.”
Look For Cloud Optimization
Merely offering a cloud service is not exactly helpful to a data center: The services must be optimized and well-suited to the needs of the organization. Raejeanne Skillern, a product manager at Intel, says optimization is key for an SME that may decide to choose cloud services meant for a much larger organization and that an SME’s homogenous environment requires customization.
“Large cloud service providers can achieve up to 10x efficiencies over traditional IT through optimized technologies,” says Skillern. “Although you may not be deploying thousands of servers like Microsoft or Google, there are CPU, platform, and software optimizations available through Intel, in collaboration with leading OEMs, that can be applied to smaller deployments to maximize cloud architecture investments through increased power efficiency, server utilization, and application optimization.”
Not All Applications Are A Good Fit
As an all-encompassing term, cloud computing often seems like an all-or-nothing prospect. Patti Dock, COO of DataMotion (www.datamotion.com), a company that provides governed integration and collaboration managed services, says that not every application is a good candidate for cloud infrastructure, especially those that require fast performance.
“Not all applications should be moved to the cloud,” says Dock. “If you are doing things where the entire process is in-house, why put the process in the cloud? If something requires millisecond response time and speed is critical, the cloud may be impractical. Then there is the issue of licensing restrictions. Or think about having someone have access to your mainframe application ‘from the cloud’—IT operations wouldn’t be too excited about that prospect.” 
by John Brandon
Bonus Tips
Ask for a contingency plan. The cloud offers key benefits, such as scalability and flexibility of services. However, the scalability is also a potential liability because a provider could go bankrupt, leaving you without the service.
“Ask your provider what the policy is for [terminated services],” says David Barley, CTO at Casdex (www.casdex.com). “For example, we have a policy that ensures that our customers will have access to their data for at least six months should anything happen to the company itself. . . . This past summer, an online data storage provider went out of business, leaving its customers without access to their data. Surprisingly, many companies have similar, nonexistent policies with regards to data storage in the event of bankruptcy.”
Barley says that the current economic crisis demands that companies take a longer look at service contracts and examine exactly how a cloud provider will retain data and make it available if financial problems lead to that company terminating services.
Lower service costs. Paul Wood, a senior analyst at MessageLabs (www.messagelabs.com), now part of Symantec, says the cloud is an opportunity for data center managers to lower overall operating costs. “By engaging with a SaaS- or cloud-based solution provider, it is also possible to look at reducing costs relating to the purchase model and the way the services are utilized, e.g. scalable, implementation-neutral, and cloud-based. This reduces the demands on maintaining internal data centers, with more focus shifting to service delivery and robust SLAs.” |
Best Tip: Consider The Programming Environment
Not all cloud platforms are the same. For example, some platforms are geared specifically for customers that use the company’s other products. “When evaluating your cloud computing options, bear in mind not only the total cost of ownership, but also the cost of adoption or migration,” says Randy Bias, vice president at cloud provider GoGrid (www.gogrid.com). “Different cloud providers have different offerings. In the case of Google App Engine, all of your code must be written in Python. In the case of Amazon Web Services, you are constrained by their networking architecture (e.g. no VPNs, loadbalancers, or advanced networking). There is quite a lot of choice in cloud providers today; however, if your product requires using a language besides Python, multicast networking, [or] VPN access or you prefer hardware load balancers, it will behoove you to look beyond the more obvious cloud offerings.” |
Best Return On Investment: Scale Services
The cloud provides an opportunity for a data center to scale services according to business needs, says Simon Crosby, the CTO at Citrix (www.citrix.com). “DCs should begin by adopting the technologies that allow clouds such as Amazon EC2 to offer both cost savings and metrics-based pricing into the enterprise environment—turning the IT department into a dynamic, cost-effective service that is responsive to the needs of the business.” |
|
|
