WP_Cloud Analytics - Mapping Requirements to Technology

3 / 54

W H I T E P A P E R
www.persistent.com
© 2017 Persistent Systems Ltd. All rights reserved. 3
1. Introduction
After having been enclosed during its lifetime behind the firewalls of an organization, the center of gravity of the
BI/analytics market is finally moving to the cloud. Recent market research surveys reveal that cloud analytics
has now reached parity with on-premise analytics in terms of adoption for new projects, with most organizations
hoping to expand their cloud implementations going forward.
An analysis jointly conducted between the Analytics practice and Corporate CTO identified three drivers that
explain why organizations are increasingly turning to the cloud for their analytics needs. First, more and more
data is in the cloud. Not only has data become available for analysis via public web services and public data
sets, but also a significant subset of an organization’s operational applications is now cloud-resident. Second,
on the demand side , the business drivers for analytics are becoming more complex. With the help of analytics,
organizations are seeking overall efficiencies and new opportunities by providing visibility into key business
processes, which explains the appetite for predictive insights driven by data from multiple sources. Established
efficiencies provided by cloud computing, e.g., operational cost reduction, also apply to cloud analytics. Third,
from the supply side, cloud platforms have reached a level of maturity where organizations can develop flexible
solutions using a large variety of data storage and management options; they provide the ability to scale
computation up or down elastically as organizations need to, and they effectively lower the total cost of ownership.
The newer capabilities available in the cloud are raising the maturity level of organizations and their approach
to analytics. At the same time, they need to evolve the processes and skills to support it. Most clients who
engage with Persistent do so because we are regarded as innovators who can help with decreasing the time to
develop and deploy their solutions. Our analysis advocates that we also should help them with a key part of the
process, namely, identifying the technical and business factors relevant for cloud analytics, and helping them
with the choice of provider and technology. In this respect, our experience in cloud analytics is key, so available
technology and best practices needs to be shared as widely as possible. This is the goal of this document,
which is written for an audience of architects, developers and technical presales people that help customers with
developing new analytics solutions or moving legacy analytics solutions and operations to the cloud.
This document is structured as follows. starts by summarizing the concerns and challenges that explain
why the industry has been slow in moving to the cloud. lists the technical and business factors that must
be considered when developing or moving an analytics solution to the cloud. We then show how to materialize
customer requirements from these factors, and we group them in categories: for instance, requirements on
the data used to derive analytics insights, on the type of queries of this data; there are other such categories.
provides technical input intended to help with mapping these customer requirements, to cloud provider
technology and services. This mapping may be used to ultimately determine the choice of provider and the
overall environment with the provider. This section is written as a series of decision points that the customer must
go through, each decision focusing on a specific area; it points out the choices to be made at each step, based
on these requirements. illustrates how we can apply these decision steps to a real customer case.
The reader should keep in mind that these steps, albeit generic, are still confined to the cloud analytics domain:
this is not about moving or developing arbitrary applications in the cloud. We assume a generic architecture with
the following mandatory components:
1. One or several “cloud databases” (a data warehouse, a Hadoop data lake, a NoSQL database, etc.),
2. Managed deployment on cloud infrastructure (at a minimum, hardware, OS, network and storage)
3. One or more processing analytic applications running on top of the cloud database: the computing power
to run analytics is in the cloud.
Finally, provides a product to product comparison of the 4 main cloud platform provider
competitors: Amazon Web Services, Google Cloud Platform, IBM Bluemix and Microsoft Azure. An
overview of the services and technologies provided by these providers is given in