Cloud Deployment Models

Cloud computing can be deployed in many ways and it all depends on the placement of
computing resources at the consumer’s location (on-premise), at cloud service provider’s
location, or at both locations. The current cloud deployment models are as follows:

  • Public cloud
  • Private cloud
  • Hybrid cloud
  • Community cloud
    Public Cloud
    This is most common deployment model. In this model, computing resources are present
    in the cloud service provider’s datacenter and are shared with various consumers in a
    multi-tenant architecture (Figure 1-2). The major advantage of this deployment model is

that you don’t need to invest in H/W and effort in setting up the cloud. The disadvantage
is that you don’t have full control of your computing resources. You can’t use this model
when there are local laws that prevent you from keeping your data outside your premises.
The public cloud model is good option for startups and any organization that wants
to avoid CAPEX costs related to DB servers.
Oracle database can be run on many public cloud providers; however the prominent
players are as follows:

  • Oracle Cloud
  • Microsoft Azure
  • Amazon Web Services (AWS)
    In the public cloud model, database related services are provided in all three service
    models (IaaS, PaaS, and SaaS). Oracle provides the following services in Oracle cloud:
  • Oracle Database Cloud Service, Virtual Image: IaaS offering to run
    Oracle database
  • Oracle Database Cloud Service: IaaS/PaaS offering to run Oracle
  • Oracle Database Exadata Cloud Service: IaaS offering for running
    Oracle on Exadata machines
  • Schema as a Service: SaaS offering to run Oracle database
    Microsoft Azure provides IaaS service where you can run Oracle database on virtual
    machines hosted in their cloud.
    AWS provides the following services for Oracle database:
  • EC2 Instance: IaaS offering
  • Relational Database Services (RDS): SaaS offering to run Oracle
    Consumer 1
    Consumer n
    Infrastructure hosted at cloud providers end
    Figure 1-2. Public cloud example
    Cloud service providers continuously add enhancements, hence it is recommended
    that you check the cloud service provider’s portal to get the latest offerings.
    Private Cloud
    In this deployment model, computing resources are placed on-premise. There is one
    more option here in which computing resources are placed at the cloud provider’s
    premises, but all of these are dedicated for consumers (Figure 1-3). The major advantage
    of this model is that you have full control of your resources and you can meet any local
    laws requiring data be kept in your datacenter. The disadvantage is related to the effort
    needed for private cloud setup.
    Database private cloud setup is done mostly in house, where companies use
    commodity server or converged infrastructures like Oracle Exadata, VCE Vblock, and
    IBM Pure app to host DB servers in a consolidated fashion. Oracle OEM 12c/13c cloud
    control is used to mimic a cloud-like setup where a self-service portal is created for quick
    provisioning and a chargeback module is used for metering and billing. In this model,
    planning, implementation, and on-going maintenance is handled by the company itself.
    Private cloud is good option for organizations that want to get ROI from their CAPEX
    investments done and where hosting in a public cloud is not an option due to compliance
    Hybrid Cloud
    This deployment model provides the best of the public and private cloud options. In
    this model, consumers use both a public cloud and private cloud to cater to different
    requirements (Figure 1-4). For example, some applications can’t move to a public cloud
    since they are running on end-of-life software, so they remain in a private cloud.
    Infrastructure hosted at customer end and cloud
    computing concepts are implemented
    Figure 1-3. Private cloud example
    This model is the future of cloud computing and world is moving toward it.
    In the hybrid cloud model, some of the databases run on-premise and some are
    hosted in a public cloud.
    The public cloud does not offer support for all database versions. You can’t run
    Oracle 10g on Oracle cloud, so you have to keep such databases in your own private
    Similarly, you might want to keep your extremely complex and mission-critical
    databases in your datacenter rather than host them in a public cloud.
    A hybrid cloud is good for situations where you can selectively move some of your
    workload into a public cloud while retaining others on-premises.