Gartner Blog Network


Planning Cloud Analytics: conceptual architecture alternatives

by Lakshmi Randall  |  July 4, 2015  |  Submit a Comment

In my previous companion blog post “Planning Cloud Analytics: Key Questions to consider”, I shared key questions to ask when planning Cloud Analytics. As a continuation, this blog illustrates sample conceptual architectures for locating cloud analytics essential components (Analytics platform components plus supporting components).  These samples are not intended to represent end-to-end architectures, nor are they intended to represent an exhaustive list of potential architectures.

Alternative architectures will result in cost variations for a cloud analytics solution. Choosing an optimal Cloud analytics architecture requires consideration of key tradeoffs described below. In addition, it requires consideration of factors such as skills, licensing, complexity, and maintenance.

             Leveraging existing investments

      Leveraging existing investments is accomplished when (1) extending on-premises analytics solution to the cloud platform, or (2) utilizing existing vendor products.

             Data Movement

      Data movement can occur between on-premises and cloud platform, as well as between cloud platforms. It is vital to evaluate and plan for frequency and volume of data movement as well as associated network bandwidth requirements.

             Acquiring best-of-breed products

      Choosing the best-of-breed products for different components of a cloud analytics solution might require evaluating product offerings supported by multiple cloud service providers (CSP).

Understanding the Sample Conceptual Architectural Diagrams

The architectural diagrams presented below depict the following:

  • A subset of essential components: The analytics platform, data integration, and analytical data stores.
  • On-premises and Cloud data sources leveraged by the cloud analytics solution.
  • Data ingress, Data egress, and Compute associated with the cloud platform for which there may be charges that typically are incorporated in subscription pricing. Charges may be allocated for cloud-to-cloud data movement.

Hybrid Cloud Solution Using a Single Cloud Platform

Placement of Components and Integration: In Figure 1, the essential components all reside in a single cloud platform. This solution integrates on-premises and cloud data sources to populate the analytical data store(s).

Data movement: Occurs between the on-premises platform and the cloud platform. Data from some on-premises data sources is consumed directly by the analytics platform.

Figure 1. Hybrid Cloud Solution Using a Single Cloud Platform

0

Cloud-Focused Solution Using a Single Cloud Platform

Placement of Components and Integration: In Figure 2, the essential components all reside in a single cloud platform. This model is suitable for “greenfield” initiatives in which there may be no preexisting on-premises data sources.

Data movement:  Leverages cloud-based data sources, thus minimizing data movement between on-premises platform and cloud platform.

Figure 2. Cloud-Focused Solution Using a Single Cloud Platform

 2

Hybrid Cloud Solution using a Single Cloud Platform with On-Premises Data integration Component.

Placement of Components and Integration: In Figure 3, the essential components are distributed between a single cloud platform (analytics platform and analytical data stores) and on-premises platform (data integration services). This model integrates on-premises data sources to populate the analytical data store(s).

Data movement: Occurs between the on-premises platform and the cloud platform.

Figure 3. Hybrid Cloud Solution using a Single Cloud Platform with On-Premises Data integration Component.

3

Hybrid Cloud Solution Using a Single Cloud Platform with On-Premises Data Integration and Analytical Data Store(s) Components

Placement of Components and Integration: In Figure 4, the essential components are distributed between a single cloud platform (analytics platform) and on-premises platform (data integration services and analytical data stores). This model leverages on-premises analytical data store(s).

Data movement: Data movement occurs between the on-premises platform and the cloud platform. Data from cloud data sources are consumed directly by the analytics platform.

Figure 4. Hybrid Cloud Solution Using a Single Cloud Platform with On-Premises Data Integration and Analytical Data Store(s) Components

 4

Hybrid Cloud Solution Using Multiple Cloud Platforms

Placement of Components and Integration: In Figure 5, the essential components are distributed between two cloud platforms; the analytical data stores and analytics platform are co-located in one cloud while the data integration services are located in the other cloud. On-premises and cloud data sources are integrated to populate the analytical data store(s).

Data movement: Occurs between the on-premises platform and the cloud platform as well as between cloud platforms.

Figure 5. Hybrid Cloud Solution Using Multiple Cloud Platforms.

 5

Some key considerations when evaluating potential architectures:

  • Co-locating all essential components (Analytics platform plus all supporting components) on a single CSP’s platform may simplify the architecture. This approach requires choosing the CSP’s product offerings, or other vendors’ products that are supported by the CSP.
  • In contrast to placing all essential components on a single CSP’s platform, distributing the components among multiple CSP platforms allows leveraging the strengths of product offerings available through each separate CSP. This approach necessitates consideration of costs associated with multiple CSPs as well as careful evaluation of architecture due to the potential for increased complexity.
  • Alternatively, distributing essential components between on-premises and (single or multiple) Cloud platforms – known as Hybrid Cloud – allows leveraging investments in pre-existing on-premises components, but requires synergy between on-premises and cloud technologies.

Category: cloud-analytics  

Tags: cloud-analytics  

Lakshmi Randall
Research Director
1 years at Gartner
15 years IT Industry

Lakshmi Randall is a Research Director at Gartner Research, focusing on data warehousing, data integration, big data and information management practices. She also follows emerging technologies such as modern data platform, cloud and NoSQL. Read Full Bio




Leave a Reply

Your email address will not be published. Required fields are marked *

Comments or opinions expressed on this blog are those of the individual contributors only, and do not necessarily represent the views of Gartner, Inc. or its management. Readers may copy and redistribute blog postings on other blogs, or otherwise for private, non-commercial or journalistic purposes, with attribution to Gartner. This content may not be used for any other purposes in any other formats or media. The content on this blog is provided on an "as-is" basis. Gartner shall not be liable for any damages whatsoever arising out of the content or use of this blog.