Companies would be wise to build their own internal clouds instead, says AS/400 founder.
Cloud computing over the Internet may have a silver exterior but a tarnished lining unless security issues surrounding the technology can be resolved, but companies today can safely build their own internal clouds that will afford far greater protection of sensitive data.
That's the assessment of Dr. Frank Soltis, widely accepted as the founder of the AS/400, who offered his thoughts on the future of 2010 trends in computing as part of a Webcast presented by Vision Solutions, "System Availability Trends 2010 with Dr. Frank Soltis" available for download from the MC Webcast Center. Soltis shared the virtual podium earlier this month with Bill Hammond, Vision's director of product marketing, who presented findings from a survey his company commissioned on upcoming IT market trends.
Soltis discussed cloud computing and parallel programming, concluding that just about everyone has a different definition of cloud computing. About the only thing they can agree on, he said, is that the Internet is the cloud. "Cloud computing is a term—it's a wonderful term—but what is it?" asks Soltis. "The first thing you find, if you go out and look at definitions, is that there are many different definitions, and, unfortunately, like a lot of these terms, it suddenly becomes a marketing buzz word."
Soltis concludes that cloud computing isn't really a new technology at all and traces similar or related initiatives over the years: grid computing, utility computing, on-demand computing, and even autonomic computing.
"My definition is that cloud computing is the newest version of service bureaus," said Soltis. Service bureaus, he explained, were a shared-resource model first popularized during the 1960s when computers were expensive, making it impossible for small and medium-size companies to afford their own systems. So they rented time on someone else's.
"If it was such a great idea back in the 1960s, what happened?" asked Soltis. IBM's introduction of the System 3 gave smaller companies a chance to buy their own computer system, he said. "You didn't have to go out and rent time on somebody else's computer." The System 3 evolved into the System 32, 34, 36, 38, and finally the AS/400 as well as even today's Power Systems.
Soltis asked the question of why service bureaus are coming back in the form of cloud computing today. He answers it by pointing out that companies are finding that the cost of managing server farms is just getting too high.
"So one of the concepts is, why don't we outsource, why don't we let someone else manage the server farms? And that really is one of the fundamental drivers for cloud computing." The other factor is the development of multi-core processors that allow for running multiple applications on a single chip.
While the need for cloud computing is there, and the technology is there, Soltis cautions that the "biggest problem today with cloud computing is really security." He recounts how there was a hacker conference underway in Malaysia at the same time he was visiting a group of System i customers there. The topic of the hacker conference was how to hack into the cloud. "It was a week-long conference for people to explain how you hack into clouds," said Soltis. "Unfortunately, as more and more companies begin to put critical information—business information—into the cloud, the first time somebody has major hacking into that and a lot of that information is stolen, there is going to be a lot of rethinking, if you will, what should be put in the cloud," he said. Soltis himself admits to storing data online—such things as family photos and music collections—but never critical information.
"Security is the biggest issue we're dealing with these days, and unless we can find some way to absolutely secure everything that's in a cloud, only certain types of applications really are going to go into cloud computing," says Soltis. "Security is still a big, big problem in cloud computing."
A more realistic approach for companies who want to use the technology is to create their own internal clouds "to be used in their business where they, in fact, can protect [the data] themselves."We're starting to see more and more of that interest—large corporations, of course—doing their own cloud computing, but [we are] also starting to see more interest in smaller systems—small and medium-sized businesses" building their own clouds, he said.
Soltis said the trend is toward server consolidation, a tactic in which companies use a larger server to create their own cloud for their own employees and customers on a platform where they can keep their data safe. Once you have that in place, however, it becomes increasingly important to have a state-of-the-art high-availability solution.
Bill Hammond noted that today's trends, based on Vision's surveys of companies, show that security is high on the list of what they will be spending their 2010 budgets on. Server hardware—likely deferred purchases from 2009—is at the top of the list, followed by data protection and storage.
"There is a general feeling that 2010 will be a better year than 2009," said Hammond. A third of the respondents to Vision's survey indicated their IT budgets would be larger than those in 2009. (Other surveys indicate nearly half of all companies have increased budgets, he noted). About half said their budgets would remain constant, and a fraction of respondents, less than a quarter, say their IT budgets will be smaller in 2010 than last year.
Hammond noted that virtualization and server consolidation are continuing trends and that cloud computing is an area where companies are exploring the feasibility of moving certain workloads off on-premise servers. Overall, availability is seen as increasing in importance as mobile devices gain broader acceptance for accessing email as well as enterprise applications, thus creating expectation of 24x7 application availability.
IT professionals continue to rely on tape backups and offsite storage as their primary data protection solution, but Hammond noted that software replication and clustering are gradually increasing in popularity among IT professionals as a primary form of data protection. Companies concerned with recovery point objectives and recovery time objectives, as well as those concerned about data loss associated with tape, should be good prospects for more-efficient high-availability solutions this year. While cost is the leading objection to implementing such a solution, Hammond noted that the return on investment often is persuasive, and today's high availability choices make HA more affordable than ever.
Other trends highlighted in the Webcast include a move toward parallel programming of applications that Soltis said is being made easier by a collection of new development tools. While standards are lacking, he said developers likely will begin to use tools now coming online to build applications that can take advantage of parallel programming capabilities built into today's hardware. While the objective is increased application speed, problems will likely arise from concurrency issues. The result is probably going to be more downtime, however, and he suggested that having a robust high-availability solution will be even more important than it was when companies were running single-threaded applications.
Readers can download the entire Webcast from the MC Webcast Center by clicking on "System Availability Trends 2010 with Dr. Frank Soltis"