GigaOm, a leading independent analyst firm, discusses the cloud data journey, best practices for deploying and using analytics data in the cloud.
Following the cautious initial adoption of cloud services a few years ago, the cloud is quickly becoming the new normal for organizations. But as it becomes the de facto solution for new analytics initiatives and migrated legacy solutions alike, it becomes increasingly evident that the path to the cloud is not as easy as it is made out to be, particularly when it comes to data. And the challenge is compounded by the proliferation of data silos in most organizations.
First, we look at how organizations should use modern data platforms such as Datameer Spotlight and Spectrum to help avoid these issues. Next, we describe best practices for deploying and using analytics data in the cloud and explore how the right third-party tool can help implement data analytics securely and at scale in the cloud. Finally, we provide generalized advice for migrating from on-premises to the cloud and provide a rationale for using a third-party toolset to enable and accelerate successful migrations.
See different strategies to bridge your data residing anywhere - on-premises, cloud, or SaaS - into the cloud for analytics.
See what challenges organizations have encountered when moving analytics and data into the cloud.
Learn about a new, agile approach to delivering analytics in the cloud that eliminates project and data risks.
GigaOm digs into Datameer Spotlight and Spectrum to show how it can make your cloud analytics more agile and successful.
In pre-cloud days, moving and transforming data was a simpler process. Origin and destination systems typically consisted of relational databases residing in an organization’s data center. A variety of robust tools based on the Extract, Transform, and Load (ETL) paradigm had been widely deployed over the years to address data integration workloads. Practitioners with ETL skills were abundant in the market. Data consumption happened mainly through software applications, Excel, or traditional business intelligence tools. And the size of data sets being transformed and moved was handled adequately, given enough on-premises storage and compute resources.
Insufficient or overly optimistic planning – Before embarking on a cloud migration of their data and associated systems, organizations need to invest time upfront to create an operating model for the target cloud environment.
Attempting a “big bang” approach to data migration – Upfront planning of the environment is vital, and an overall strategic sense of purpose and vision is a great motivator for staff that can help unlock the budget for the initiative.
Underestimating the complexity of migration – Reviewing and pruning data at this stage might be a wise step, but it is not applicable for every organization. Storage and processing costs for the migration need to be planned.
Not planning for differences in target architecture – There is no like-for-like mapping between on-premises Hadoop infrastructure and cloud-native solutions from the public cloud providers. Organizations have fundamental choices to make here.
Inadequate emphasis on data governance and security – Migration offers an opportunity for organizations to shore up their existing data governance processes. Ideally, data governance is already operating on the on-premises data but the cloud enhances and complicates the need for it.
Neglecting organizational buy-in and staff skills realignment – Examining existing staff skills and retraining, hiring, or changing the skills mix is vital to success, as not all on-premises skills map directly to the cloud.