Header Ziff Davis
Advertisement
Advertisement
Wednesday, June 11, 2008 5:17 PM/EST

IT Management in the Age of Cloud Computing

One of the questions that becomes apparent when thinking about the advent of cloud computing in the enterprise is how IT organizations will go about holistically managing everything inside and outside the cloud. After all, not all the on-premise software being used today is going to magically transfer itself to a service in the cloud. That means that as far as anybody can see into the future, IT organizations will need visibility across any number of clouds that are being used to supplement their internal IT operations.
Worse yet, the advent of virtualization is redefining the way we think of managing applications across dynamic sets of virtual machines that essentially mask the underlying physical servers. That means that not only do IT organizations need a way to manage assets in the cloud, they also need tools that manage both virtual and physical servers.
One company that is creating a new approach to IT management that reflects all these concerns is Hyperic based in San Francisco. The fundamental concept behind the Hyperic approach is that no assumptions can be made concerning what platforms and applications will need to be managed. To that end, Hyperic has created an agent framework that can easily extended to not only any given application or system, but also any service that has an open application programming interface.
That means that by leveraging the Hyperic framework to manage any number of IT assets in a way that maintains maximum flexibility for the IT organization. And has demands to support new technologies that drive innovative business models continue to grow rapidly, it's pretty clear that IT organizations need a new approach to systems management. That approach is ultimately up to the individual IT organization but what's intriguing about the Hyperic approach is that it points they way to rethinking how we approach systems management.
Trying to restrict the number of products and technologies that make up an enterprise is usually a losing proposition because invariable competitors embrace new technology to change the business landscape. At the same time, struggling to support an infinite number of products that come with their own management systems is nothing short of corporate suicide. The ultimate systems management challenge facing everyone today is how to inject enough flexibility into a structured approach to systems management that limits the downside effects of the chaotic nature of enterprise computing today.
Of course, many an IT organization has tried to answer this challenge by rolling their own set of management tools so it's nice to see somebody in the vendor community trying to step up to the problem. And who knows, tools like these may even enable IT organization to build their own set of cloud services as opposed to seeing every thing outsourced simply because a third-party service simply has a better set of tools.
Whatever ultimately happens, limiting the productivity gains to be derived from new technologies and computing models simply because we don't have the tools we need to manage them effectively is a paradox that we can no longer afford.

TrackBack

TrackBack

http://blogs.eweek.com/cgi-bin/mte/mt-tb.cgi/13931

Comments (2)

Michael makes several good points about the need for new tools to handle the upcoming evolution to utility / cloud computing. I think one important point is being missed here: even when we have the right management tools, who will do the managing?

IT staff right now have a great set of skills that allow them to deal with today's reality of boxes and operating systems. What they don't have is the new set of skills that they are going to need to deal with the brave new world of cloud computing. The big question here is who will do this and how will they learn to do it.

Fundamental questions like what apps are best suited to move to a cloud, how do we choose the right cloud to use, what metrics are the best to judge cloud performance, and finally, when we grow dissatisfied with a cloud just how do we move from one to another?

There are no easy answers to these questions. However, having the right cloud management tools should be seen as a first step in trying to develop the new IT skills that will be needed.


- Dr. Jim Anderson
Blue Elephant Consulting
www.blueelephantconsulting.com

Certainly cloud computing and virtualization are complicating the performance and availability management of mission critical applications and services. It is interesting that you called out Hyperic for their adapter framework, however, any management/monitoring solution worth its salt has a similar toolkit for creating integrations to new technologies. But once again, Hyperic is based on the same approach as traditional monitoring tools - setting hard thresholds for different metrics, getting tons of alerts and then trying to sift through them and figure out which alerts are applicable to the current problem and which are not.

What needs to change to avoid this is to take a new approach to systems management. One that uses sophisticated real time analytics to learn the normal behavior of the infrastructure components that make up an application/service and only alerts to abnormal precursors to problems. Such a solution should also be able to use advanced correlation techniques to determine how business performance and user experience key indicators are affected by infrastructure components. With this type of approach, false positive alerts are eliminated and when problems occur, root cause can easily be determined. Additionally, this approach allows a proactive approach to systems management because the precusors to problems are captured and problems can then be predicted in advance if they begin to re-occur. The approach I described is available today and can reduce a tremendous amount of manual effort. This can allow an Operations team to scale in a more complex environment without requiring additional labor spend. Considering that 70% of IT budgets are consumed with labor costs just to keep the lights on, this is a significant capability that all Operations groups should be considering.

Post a Comment

 
 


Advertisement
Advertisement