[caption id="attachment_913" align="aligncenter" width="518" caption="Many companies are trying to make cloud computing fit their old model of IT management."] [/caption] Most enterprise IT organizations are divided into stacks of computing, anchored on specific computing platforms. Application servers generally run on Windows or Linux systems, databases run on Unix servers and applications that require heavy-duty transaction processing run on the mainframe. And it’s pretty much been that way for the past three decades. Many companies are adapting their IT infrastructure for cloud computing in ways that mimic this aging model. In this “evolved” setup, some stacks of computing run on private cloud-computing platforms managed by in-house IT staff. Meanwhile, other stacks run on private clouds managed by third-party cloud service providers, with some applications running on shared public-cloud infrastructure. It’s pretty clear that IT will need to orchestrate the management of application workloads across all those instances. In fact, companies such as Accenture are already gearing up to manage all those instances of cloud computing on behalf of their clients. Accenture recently inked an alliance with Microsoft under which the former will manage application workloads on behalf of its clients running on the Microsoft Azure cloud platform. According to Paul Daugherty, chief technology officer for Accenture, the alliance is one of the first of many such arrangements: Accenture envisions helping customers manage a federated set of cloud computing services that will run on premise and in data centers managed by a variety of providers: “We expect to soon have a series of alliances in place with cloud service providers that will allow that to happen.” In theory, cloud computing should transform the way we manage IT. Instead of trying to cope with islands of IT infrastructure that need to managed in isolation, organizations can leverage the cloud to unify the management of application workloads, regardless of where those workloads actually run. But in the short term, it looks as if many organizations have decided to manage cloud-computing assets in a manner similar to most IT infrastructure today: isolated and siloed. Most platform-as-a-service (PaaS) offerings, for example, can only support applications written in only one or two programming languages. As a result, application workloads are likely to remain loosely coupled—one reason that companies such as Eucalyptus Systems have chosen to focus on optimizing the management experience between private clouds and a single public-cloud environment. In the case of Eucalyptus, that public cloud happens to be Amazon Web Services (WAS). Eucalyptus CEO Marten Mikos believes it could be years before we see workloads federated across multiple cloud providers. From his perspective, the optimal course involves concentrating on bridging the divide between private clouds and Amazon, as opposed to worrying about building so-called cloud operating systems such as OpenStack. “A lot of the alliances such as OpenStack are really just anti-Amazon moves,” he noted, “that are based on assumptions about the usage of cloud computing that may not reflect any reality for years to come.” Industry analysts such as Jeff Kaplan, managing director for ThinkStrategies, an IT consulting firm specializing in cloud computing technologies, suggests that, thanks to a significant customer requirement, management tools will quickly evolve to meet the federated needs of the cloud. “I just don’t think it will necessarily be the leasing players in enterprise IT management today that are necessarily going to be the first to develop that capability,” he said. “They can’t federate the management of IT inside the enterprise today, never mind what’s happening in the cloud.” While there’s no doubt that cloud computing will reshape the way enterprise IT is managed, it’s probably going to take a lot longer than most people think. In the meantime, expect that IT organizations will continue to manage computing stacks in a business-as-usual way, whether the datacenters in question are on-premises or in the cloud. Image: Jakub Pavlinec/Shutterstock.com