For those of you that follow my blog and Twitter account, I recently took a new role with HP. Now in addition to helping customers with blade servers, I’m taking a focus on cloud solutions. While I hate the term cloud as much as the next guy, but I know the concepts are solid and they represent the future of our industry. So while I’m starting my journey to the cloud, I thought I’d share one of my client’s journeys as well.
Last week I had an interesting discussion with this client about how they can take advantage of cloud. They were just starting to look at alternatives and trying to build a roadmap for the future. Their thinking was to use cloud concepts to streamline their hardware and OS deployment. Their initial scope was to look for a solution to deploy Windows and Linux virtual machines, and possibly Solaris. One of the first questions that come up was about using the Self Service Portal. They couldn’t see why a traditional IT shop would want to deploy servers on demand, never mind thinking about applications on demand.
This highlights the first issue that must be addressed when looking at cloud solutions. Cloud isn’t a special set of hardware. Cloud isn’t about a new whiz-bang software product. At its core cloud is a different way of thinking about how IT solutions are designed and delivered.
Instead of deploying large monolithic scale-up applications, cloud encourages modular scale-out applications. Instead of sizing an application environment for a worst case scenario, cloud encourages right sizing the application. If the sizing is wrong, or demand grows or shrinks, nodes can be quickly added or removed. This will optimize power, licensing and management costs as well as CapEx.
This is just one are that cloud can help. Additional use cases include:
- Standing up QA/test environments on demand to improve application quality.
- Scaling up and down applications to meet seasonal demand
- Bursting to meet unplanned demand. We’ve all heard horror stories of web sites crashing when a product launch is more successful than planned. We need to make sure IT doesn’t get in the way of business.
After we discussed the use cases they needed to consider, we talked about what they needed to deploy. The initial requirement was to deploy hardware and operating systems. This is commonly referred to as Infrastructure as a Service (IaaS). While IaaS has value and is a good starting point for organizations looking at cloud, it’s usually not the ideal end state.
To fully realize the benefits of the use cases described above, the solution also needs to deploy applications, and manage them once they are deployed. Many of these application environments require more than just virtual machines. In many cases physical servers and enterprise UNIX are also required.
For this client it is likely their move to the cloud will be a journey that might last several years. They don’t want to move all at once. So they are likely to be running cloud and traditional IT side by side for years to come. This brings home a few important points.
Pick components that work towards the desired end state. There are many tools that will meet the requirement of an intermediate step, IaaS as an example, but are not part of a larger portfolio to satisfy the long term requirements for PaaS or SaaS solutions.
Pick tools that work in cloud and traditional IT environments. Pick tools that work with all components of the infrastructure, not just the components that are used in the current phase. This will reduce cost and complexity over time.
Every Journey to the cloud is different. What will yours look like?
