10. ◊ Identity (Keystone): Keystone provides identity authorization and
authentication service to all the OpenStack users.
◊ Images (Glance): Glance provides image services to the VM
instance. Glance discover, register and deliver services to image
servers and disks from image store via a Restful API (i.e. Glance-
API) communicating with compute service.
◊ Dashboard (Horizon): Horizon provides a graphical user interface
access to the system administrators and users to the provision,
control, and management of OpenStack cloud resources. An
alternative to the horizon is the CLI access and access via third
party tools.
11. 1. Dashboard/CLI/third party API through the restful service
receives user request and forward it Nova-API.
2. Nova-API interacts with Nova-DB, which then creates the
database (DB) entry for the instance.
3. Nova-API sends the Remote procedure call (RPC) request to
the Nova-Scheduler through Message-queue (which is visible
through RabbitMQ management interface.
4. Nova-scheduler interacts with Nova-DB to find the available and
suitable host (the process is mostly achieved by going through
complex scheduling and VM allocation algorithm [25, 26] by
weighing and filtering).
5. Nova-scheduler sends the RPC call request to the Nova
compute for launching instance to the appropriate host. Note
that if the host is an external system, then communication takes
place to the compute nodes (c.f., Figure 3).
6. 6. Nova-compute sends a message to nova-conductor to fetch
detail instance information such as host-id, flavor, specification
(RAM, CPU, Disk).
7. 7. Nova conductor interacts with Nova-DB through messaging
queue and makes a RESTful call by passing the token to the
Glance-API to get image data (URI, name, and size among
others).
8. 8. Nova-compute also passes Restful token to Neutron through
Neutron-API to configure Neutron based on user network
specification including IP address
9. 9. Nova-compute also passes the similar Restful token to Cinder
through Cinder-API for volume based on user request to attach
storage volume.
10. 10. Nova-compute generates required data for hypervisor driver
and executes the request on the hypervisor. Further processes
leading to the launching of VM will continue at the hypervisor
level which may need further operations from host operating
12. The proposed analytical model is based
on the OpenStack IaaS Cloud computing
architecture. It comprises of three
network types: Public network, Admin
network and high-speed data network
between computing resources. It also
comprises of Controller nodes i.e.
gathering requirements for VM instance
creation and similar services). Compute
nodes (which host the VMs instances
created for the compute service), Admin
Node (responsible for managing the
hardware i.e. booting and shutting down
as needed),
Storage Nodes (responsible for
housing/hosting the VM instance
volumes, i.e., Hard Disk Drive (HDD)).
Proposed Model
13.
14.
15.
16.
17.
18.
19.
20. Note that the values of the mean external arrival rate are set from 0.1 to 2.0 with an
interval of 0.1 amounting into 20 sample values.
21.
22. Bottleneck Analysis : .It can be observed that for every value of
ߣ(lambda) ,the Station 2: Controller has higher utilizations than
those of Station 1: Firewall and Station 5: Storage.
As these stations begin to reach higher utilizations values for more
extreme values of ߣ(Lambda) it is noted that the Station 2:
Controller has the highest utilization equal to 0.9 whilst the Station
1: Firewall and Station 5: Storage reach their highest utilizations at
0.8 and 0.7 respectively.
From these typical numerical experiments, it is implied that the
Station 2: Controller (Station 2) is the bottleneck station of the
open QNM and thus, it is the one that imposes limits in the
performance levels of the CCP.