At Peak 10, we’ve had customers in industries such as healthcare, finance and logistics approach us for guidance in leveraging cloud services to support a big data initiative. Their first challenge is always the same: What goes to the cloud — the whole project or just production?
Depending on the nature of the initiative, we recommend starting with an assessment of the condition of the data. Is it structured or unstructured? What does putting the front-end servers into the cloud looks like and dedicating the now-free existing physical servers for the project itself?
Next, it’s a great idea to compare costs. What would it cost to run testing and development in the cloud compared to purchasing additional servers? Can the data mining and data reductions benefit from the elasticity of the cloud?
As they proceed from the proof of concept phase to production, many customers find it best to move to a hybrid IT solution consisting of cloud services and physical servers. Development and QA can benefit from the flexibility and cost benefits that cloud computing offers over time. Applications may have specific clustering configurations on top of the server infrastructure due to the amount of data included in the project. Based on the specific project needs, the data may also warrant physical servers to process the data while user interface (UI) is presented by cloud servers. In addition, meeting HIPAA, PCI and other regulatory concerns may require the use of dedicated, physical servers for certain tasks.
Many of our big data customers have also discovered that the value of leveraging a hybrid IT strategy is on front-end servers, rather than housing their initiatives long term. For example, a healthcare company wanted to produce a proprietary product based on data they had collected on procedure metrics, insurance claims and other reporting over the years. However, they couldn’t efficiently crunch the numbers due to the volume of data they had without a building out new infrastructure resources.
The cloud didn’t make sense to house the data, given limitations on access and the cost. They determined that they would be better off purchasing servers. But they still needed a way to present that data and take in new data from mobile devices and remote clinics.
Peak 10 offered a cloud-based solution for their testing and development that used only six servers. When the proof of concept was completed, we moved their data to a private cloud with colocation to protect it. Meanwhile, they moved their front end to the Peak 10 cloud. The resulting hybrid solution provided them with security and scalability, and they didn’t have to focus on running that part of their infrastructure.
If you’re considering a big data initiative, you may find that workloads such as web servers, reporting servers, front-end servers or even middleware are great candidates to migrate into a hybrid IT solution. This will allow you to control capital spending on hardware investments — and help you better meet the specific needs that arise when working with big data.
For more information on hybrid IT, take advantage of this free resource: Hybrid IT: Make Colocation Your Conduit to the Cloud.