Dan's Take

Cloud Computing, Then and Now

Many things have changed -- and not changed -- since 2009.

• By Dan Kusnetzky
• 07/14/2016

I don't know what came over me. I found myself reviewing some notes from an interview with John Suit, CTO of a now defunct company, Fortisphere. He was discussing his projections of the key trends in management of cloud computing environments and applications that he expected to unfold in 2010. Although an ancient interview, it stuck me how little things have changed -- and how much things have changed in other areas.

Here's what Suit had to say back in 2009:

"Everyone talks about the move to cloud computing, and have for years. In fact, in 2009 virtual administrators realized that in order to get to the cloud, they needed to first establish a level of service for their application owners and then fully understand and monitor the components of that service, in a way that directly translates to the cloud. They also began to realize that if they had any hope at all of migrating to the cloud, they must first get their houses in order.

The tolerance for a lack of real management solutions in virtualization is over.

In 2010, service management will be the key to taking virtualization to the next level. Service management will emerge as a key to help IT organizations gain a complete understanding of the services they provide, what the components of the service group are comprised of, and the dependencies and relationships that must be maintained to provide the required level of service.  IT directors will seek out solutions that can be used to easily migrate their service groups within their infrastructure, and will be able to facilitate a transition to the cloud.

In 2010, multi-platform management will be critical. Some vendors understand that dependencies are important, but they too often develop solutions that require dependency of specific frameworks. A solution that only understands VMware or only Hyper-V is going to have a hard time providing any real information about a specific service group in the cloud, when those framework specific interfaces are no longer available. Organizations will begin to see that dependency mapping is not enough; they must also have the communication relationships of those infrastructure components, as well as monitored state that is based on thresholds that are acceptable to the IT administrator and the application owners."

At the time, I pointed out that comprehensive management and security solutions are always areas that are given little thought by many organizations. This is true today, even though there are many highly-publicized failures in both areas.

Although industry experts suggest that both of these functions be "baked in" during initial planning and design, the focus of most organizations is still tactical rather than strategic in nature. Many still focus on managing specific applications and application components, and don't rise above the mere functional view to consider managing their entire multi-platform, multi-workload, multi-vendor environment.

The 2009 Dream of Single Console Management
At the time, I thought that a comprehensive management architecture was very hard to design and implement. Since that time, we've seen a dramatic fragmentation of applications into services and micro-services that may execute on multiple platforms, both physical and virtual, in multiple datacenters, both operated by the enterprise and also by cloud service providers. That makes the creation of such a comprehensive management environment even more difficult.

Back in 2009 -- and still today -- many enterprises, lacking a comprehensive view, purchase point solutions that deal with the issues of one platform or one workload, and don't adopt solutions that will allow everything, everywhere to be managed from a single console.

I still find examples of management suppliers offering a product that can deal with the needs of x86-based systems running a single operating system and call it a "datacenter management" product. This, despite the knowledge that most datacenters of any size contain mainframes; midrange systems running UNIX or some single-vendor operating system; industry-standard systems running some combination of Windows, Linux and, yes, other operating systems, such as UNIX. They also have intelligent storage systems, networking gear; and, possibility, intelligent coffee makers.

Since these suppliers don't have products to address such a diverse set of needs, they just ignore them in their public statements (maybe hoping that they'll die of neglect.) Instead, they wave their arms and claim to be able to manage the datacenter.

Dan's Take: Days of Future Past
Although it would have been very nice if 2010 was the year that organizations elevated their viewpoint to look across all their datacenter(s) and think about management and security, it just didn't happen that way. In fact, it still hasn't happened in many enterprises. Although there are more products that have the breadth and depth of capabilities necessary to instrument, gather data, analyze that data and provide clear guidance for operations staff, there are still many examples of single-focus products.

Back in 2009, cloud computing was just emerging, and there was a wide industry debate over what it really was. Some suppliers asserted that everything they were doing was somehow a form of cloud computing. Others were much more specific. At that time, it was my view that cloud computing was (and still is) just an outgrowth of long-established outsourcing practices riding on the back of colocation and managed services platforms.

The issues of managing a combination of applications, application frameworks, databases and other IT components that reside in enterprise and service provider datacenters are still here, all these years later.

I guess that everything old is new again.