Cloud Analytics Migration: Go With The Need
The Cloud offers access to new analytics capabilities, tools, and ecosystems that can be harnessed quickly to test, pilot, and roll out new offerings.
Join the DZone community and get the full member experience.Join For Free
The Cloud offers access to new analytics capabilities, tools, and ecosystems that can be harnessed quickly to test, pilot, and roll out new offerings. However, despite compelling imperatives, businesses are concerned as they move their analytics to the Cloud. Organizations are looking at service providers who can help them allocate resources and integrate business processes to boost performance, contain cost, and implement compliance across on-premise private and public cloud environments.
The most cited benefit of running analytics in the Cloud is increased agility. With computing resources and new tools available on-demand, analytics applications and infrastructure can be developed, deployed, and scaled up — or down — much more rapidly than can typically be done on-premises.
Unsurprisingly, cost reduction is seen as a significant benefit of cloud-based analytics. A complex algorithm processing large volumes of data may require thousands of CPUs and days of computing time, which can be prohibitive for companies without existing in-house compute and storage resources.
With the Cloud, organizations can rapidly access the required compute and storage power on demand and only pay for what they use. Research shows that migrating analytics to the Cloud can double an organization’s return on investment (ROI).
Standardization, cited as the third most crucial driver of migrating analytics to the Cloud, is strongly linked to the first two benefits of increased agility and reduced IT costs. Also, standardization helps organizations with simplified, streamlined IT management and shortened development cycles.
The Cloud offers access to new analytics capabilities, tools, and ecosystems that can be harnessed quickly to test, pilot, and roll out new offerings. For instance, organizations can take advantage of cloud-based data integration and preparation platforms with pre-built industry models. Leverage cloud services that offer powerful graphics processing unit (GPU)-based compute resources for complex analytics and tap into a collaborative ecosystem of data analysts within a federated data environment.
Cloud Analytics Migration — Go With The Need
There are multiple migration strategies available and depend on your needs and goals. Some of them fast, others slower. Usually, the migration process includes five phases:
- Evaluate Opportunity — analyze the cost and benefits associated with migration to the Cloud.
- Discover and Analyze — access cloud migration portfolio and formulate a migration plan.
- Plan and Design.
- Migrate, Integrate and Validate.
- Operate and Optimize.
Cloud Analytics Migration Strategies
There are six commonly used strategies for analytics migration to the Cloud:
- Lift and Shift.
- Lift and Reshape.
- Drop and Shop.
- Re-write/Decouple Applications.
- Retain/Not Moving.
It is important to note that most migration projects employ multiple strategies, and there are different tools available for each plan. The migration strategy will influence the time it takes to migrate and how they are grouped for the migration process.
Lift and Shift
We lift our application and shift to the Cloud. This strategy is fast, predictable, repeatable, and economical.
Lift and Reshape
This approach is similar to the previous one. But you will also deploy the last version of the software.
Drop and Shop
This approach allows you to replace your application with a new one.
This approach is about changing application binaries before migrating to the Cloud. This could apply to the custom and open-source solutions.
This approach will decommission your application on-premise.
This approach allows you to leave your solution on-premise.
Concerns With Migration
Despite compelling imperatives, businesses are concerned as they move their analytics to the Cloud:
Organizations perceive data in the Cloud as being inherently less secure than data on-premises. This perception may be the result of sensational news coverage on cybersecurity hacks on cloud platforms. In reality, on-premises infrastructure is no more secure than well-built cloud systems. The reverse is more likely to be accurate since established cloud providers comply with the most stringent security requirements and invest heavily in security solutions, personnel, and resources.
Selecting the Right Architecture and Infrastructure
Organizations are concerned that choosing the wrong platforms for their analytics applications could lead to performance issues, data fragmentation, integration challenges, and vendor lock-in further down the road.
Integrating Existing Applications With Newer Cloud-based Applications
Organizations often run diverse applications built on complex interdependencies, while data is housed in various silos and different formats. Without accurate data for mapping application dependencies, resources, and resource utilization, respondents will see integrating existing applications with cloud-based ones as a significant challenge.
Data Management and Governance
Given the diversity of data types and sources, organizations must grapple with data access and governance and increase regulatory oversight on how the data is managed and where it resides. These concerns may give organizations pause when they consider moving their data to the Cloud.
The Most Formidable Barriers To Operationalizing Analytics In The Cloud
Difficulty in Deploying Models Into Business Processes and Applications
This may arise from a combination of low data quality, the lack of alignment between data team and business units, or models that are too complex for users to understand or use.
Lack of Accurate Data and Analytics Governance
Simply throwing time and money in data preparation and model building across different parts of the enterprise is not enough. Without appropriate governance, such uncoordinated efforts may not be aligned with business objectives or realities and, therefore, not worthwhile to begin with.
Data Privacy Concerns
Regulations like GDPR have made using personal data for analytics much more challenging. In most cases, Asia Pacific organizations dealing with customers that fall within the scope of EU regulations need to obtain customer consent or anonymize the data before using them for analytics — resulting in increased effort and cost.
Lack of Relevant Skills or Staff
For many organizations, finding skilled personnel like data scientists capable of finding, organizing, and interpreting data with the right tools can be a real challenge. This problem is not expected to be resolved anytime soon.
Unable to Address Data Quality and Data Preparation
Data quality and preparation issues generally result from disparate processes, fragmented systems, or siloed information stores. What’s more, as data volumes continue to grow exponentially, organizations can find it hard to stay ahead of the problem.
Published at DZone with permission of Rahul Asthana. See the original article here.
Opinions expressed by DZone contributors are their own.