Cloud Changes Everything, Part 4 – The Changing Structure of IT Operations

[paulstrong]

The externally driven change in role is matched by an internal transformation in terms of how IT organizations are structured.  The traditional data center is one of silos of control. The very nature of the technology that comprises IT infrastructure, and the lack of management tooling that has been able to treat it systemically or holistically and hide complexity, has forced the creation of expert groups, each with a specific focus on one aspect of the infrastructure, such as networking, storage, compute, operating systems, databases, applications and so forth.  This “divide and conquer” approach allowed organizations to deal with the complexity of the technology and it’s deployment, to some degree.  But at a high cost!  Specifically the creation of organizational silos, that inherently tend to toward being somewhat introverted and focused on their technology rather than on the business services they effectively run.  These silos naturally become a barrier to both efficiency and to agility, as the processes required to effect change tend to be implemented as ticketing systems where tickets are either passed on like relay batons or cascade as the workflow moves from silo to silo.  The net result, if care is not taken, is a set of inward looking organizational silos, goaled with the most basic of service levels, usually expressed in the form of tickets, time to resolve tickets, number of tickets resolved and so forth.  This is at odds with the desires of the business for service management.

Virtualization technologies have started to change this.  Whilst not directly addressing the ticketing issue per se, they have started to blur, and, in some organizations, dismantle the silos, increasing agility and shifting the focus more on the application or service.

The adoption of server virtualization, in general, and VMware specifically, was heavily driven by two desires.

The first was the obvious one, to maximize the utilization of expensive resources, i.e. servers.  By having multiple, compatible workloads sharing a physical server, greater efficiencies are realized.  Hosting multiple operating systems, each hosting it’s own workload, is particularly attractive because of the effective isolation of workloads and because most workloads still tend to have specific requirements of the underlying operating system, including patches, tuned parameters, versions of specific software components, e.g. DLLs, and so forth.

In effect the virtual machine becomes the method for packaging workloads/applications, and this leads to the ability to meet another strong desire, that of simple application provisioning.  Packaging or encapsulating applications within virtual machines solved the industry wide dilemma of how to provide the basis for reliable automated provisioning.  The ideal way would have been to have an agreed upon, i.e. standard, way of describing software manifests and standard methods for provisioning, that multiple software tools could thus implement, providing interoperability and coverage of all applications. The technology industry was never able to achieve this as a collective, but virtualization effectively did an end run around the issue.  Whilst in some sense not perfect, it is more than good enough, which is what actually matters.  And indeed this kind of encapsulation affords many more opportunities in terms of what can be done to manage

applications.

In any case, the ability to meet these needs has made the adoption of server virtualization extremely compelling.  The end results of virtualization, i.e. cost saving and agility, combined with tools that are simple to use, has resulted in an incredible rate of adoption.  As the tools have become richer and because by their nature, i.e. they enable aspects of networking and storage to be managed in an integrated fashion, they have become the path of least resistance in terms of getting certain tasks done.  Tasks that used to span several domains or silos are now being undertaken by the virtualization admin using a set of tools that present a systemic view of infrastructure.  As the boundaries between the silos become blurred, more and more opportunities for process streamlining and automation, are presenting

themselves.  And as each is taken, agility and efficiency benefits are accrued.

Next… The Perfect Storm