Moving Big Data analytics to cloud is cool, but are you ready for it? Here are some best practices that will ensure that the transition is smooth and successful.
Big Data analytics in the cloud can open a whole new world to your business. It can help you scale limitlessly and offer analytic capabilities to your users on an anytime, anywhere basis. However, if you really want to capitalize on the benefits of cloud, you need to deploy analytics in a way that it is easily available, fast, and reliable.
BI on Big Data in the cloud can fail for a number of reasons — ranging from lack of coordination between tools, poor adoption by users, inability to meet the expectations of the business users, and many others. With the right amount of research and planning, you can ensure that your cloud implementation project finishes on time, within budget, and with high quality results. I’ve put together a best practices guide that will help you assess your readiness and prepare yourself for the move.
Best Practice 1: Assess the impact of data movement
The first step of this project is to migrate your data to the cloud platform. Before choosing the service provider and the subscription, you need to assess the current state of your data and the impact of data movement.
- What types of data do you have – streaming data, file systems, etc? Is the data ready to move? If not, how much effort will be required to make it ready-to-move?
- How easy or difficult is it to move your data? Can it be moved in entirety? What is the cost involved?
- Is your cloud platform ready for your data? Have data security concerns been addressed? Does it integrate with your on premise IT?
- What kind of tools and services do you need to access your cloud data? Does your existing BI solution support the move to the cloud? Is performance affected?
The answers to these questions will help you assess the cloud services you require and the time, effort, and money that would be required to move the data.
Best Practice 2: Get clear on BI expectations of the business users
Data is meant to solve real-life business problems, irrespective of whether it resides in the cloud or in an on premise environment. Therefore, evaluate all tools, platforms, and processes from the business perspective.
Ease of use, self-service, high availability, and minimum wait times, should be major concerns while building your BI platform on the cloud. Access to Big Data should be seamless and the users should not be bothered about the size of their data or its security.
Before starting the implementation, it is important to identify the business users who would be using the BI solution and get clear on their expectations.
- What kind of reports do they need? Are the reports fixed or ad-hoc?
- How quickly do they need their answers?
- Are they open to working in a new environment? How much time would they need to transition to the new technology?
- How much IT intervention or hand-holding do they expect?
If the business users expect response times in seconds but your BI solution is taking minutes and hours to fetch results from the cloud, you may have to rethink. Heavy-lifting of cloud data for analysis would be an expensive and undesirable overhead. Reconsider and adjust your solution to meet the expectations of the business users.
Best Practice 3: Create a blueprint for an end-to-end solution
The relationship between BI tools and your cloud platform is a symbiotic one. They both need to work hand-in-hand to be able to deliver results and score on performance. Just as cloud platforms offer the inherent capacity to accommodate your growing needs, your BI solution should also be able to scale up to meet the changing requirements without compromising on performance or speed. It should support the elasticity and agility of your cloud platform. Besides this, the cost of accessing cloud data should not dilute the cost benefit of the cloud storage. This becomes an important consideration if you are dealing with massive volumes of data.
Moving to the cloud is a big initiative that requires investment in terms of time, people, and processes. Therefore, it is important to create a blueprint for an end-to-end solution. You should choose technologies that work in synchronization to take care of everything from data collection and storage to actual consumption of the data by the business users.
Best Practice 4: Follow a phased approach
The approach that works best for you depends upon the nature and scale of your data but irrespective of that, you should focus on a solution that supports easy transition with minimum disruption to your existing infrastructure.
A well-balanced, phased approach will lead to success. Your data may be huge and serve several use cases across the organization, but you can start by moving it in phases. A hybrid environment that consists of your on premise legacy data warehouses and your cloud Big Data platform with a unified BI consumption layer that has the capability to handle data in both the environments, would be the perfect solution. You can also start with a single use case and a smaller team. Take each use case to its logical end where business value is clearly visible and then start on the next use case.
Many of the concerns mentioned above can be solved by choosing the appropriate tools and platforms. OLAP on Big Data is a promising technology that offers the benefits of speed and scale. The flexible computing capacity of the cloud environment makes per-processing of Big Data cheaper and easier. A well implemented MOLAP solution supports all major BI tools, enabling easy access of cloud data. Scaling up gets cheaper and it becomes easy to add fresh data, more dimensions, and new data sources. In my opinion, the key feature that works in favor of an OLAP solution is its performance. It can serve almost any query instantly, making interactive access of Big Data on the cloud a reality. When BI becomes fun, ROI is a natural outcome.