W H I T E P A P E R
www.persistent.com
© 2017 Persistent Systems Ltd. All rights reserved.
6
We have chosen to expose performance and scalability also as separate dimensions belonging to the non-
functional requirements category. We distinguish performance levels as being Very High, High, Medium or
Average, as well as types of scalability such as Scale Up, Scale Out and Elasticity. Please refer to Appendix 1
for a discussion about what is meant by these terms. Our performance scale will be applied informally
throughout this document, for the most part ignoring workloads (so it should be taken as a very rough
indicator).
Customer requirements must be mapped to cloud platform services (see below), and a choice of provider.
4. Decisionpoints for selectingcloud technology, services andprovider
Whatever the drivers may be, whether cost reduction, IT staffing difficulties, or business drivers (e.g., enhancing
business processes or improving customer experience), we assume in this section that the organization that
Persistent (PSL) is servicing is convinced of the benefits of moving its analytics operations to the cloud, or
developing a native cloud analytics solution.
This section provides technical input intended to help with the process of mapping the customer requirements
to cloud technology, services and platform provider. It is written as a series of decision points that the customer
must go through. Each decision focuses on a specific area, points out the choices to be made at each step, and
considers the relevant requirement dimensions out of the 20 dimensions introduced in the previous section. We
will be referring to this organization as “the customer” (we might also be referring to it in second person, as in the
expression “your requirements”).
4.1 The type of cloud (public, private or hybrid)
The first decision enterprises must make when moving data to the cloud is to choose the right environment. While
public, privat
e 1and hybrid clouds all have benefits, determining which model best meets their needs is a crucial
step on the path to the cloud (both migrations and new deployments). Organizations must make this decision
based on various dimensions described below:
a.
Security
and
compliance
– As mentioned in
section 2 ,these are the top two obstacles for moving
the data to the cloud. At the same time, security has also been recently identified in
[1]as a technical
driver for adoption of cloud analytics. Indeed, as cloud platforms mature, fear of security issues have
lessened: products from platform vendors such as Azure SQL Data Warehouse, Google BigQuery,
Amazon Redshift as well as other infrastructure offerings (e.g., Virtual Public Networks) have swaths of
security features to guarantee the safety of customer data at every point in its journey.
Cloud Platform
Software as a Service (end user application)
Platform as a Service
Infrastructure as a service
Private cloud (virtualized hardware)
Database as a Service
Data Movement as a Service
Security services
Development services
Event-driven services
Message Queue services
Application Resource Management services
Monitoring services
Archival services
Storage as a service
Virtual machines
Network
Analytics
BI as a service
ML as a service
Externally managed
On premise (no management)
Additional PaaS services
1
It should be understood here that we refer to a dedicated deployment (thus, for a single tenant) on virtualized hardware managed by an external provider –otherwise we are
talking simply about an on-premises deployment.