Patterns Of Cloud Infrastructure

Talks about cloud computing and cloud hosting are the trend. Through virtualization, these conversations are becoming a reality in many companies (private clouds) and service providers (public cloud). These patterns formalize certain rules and consider the problems and solutions that arise in the process of reflection on various aspects of cloud infrastructure.

Something very interesting:

Template virtual machine: The easiest way to standardize the deployment of virtual machines in a virtual infrastructure that meets certain levels of security and contains the necessary application. Here we consider the problem of the hierarchy of templates as an enterprise application typically being implemented on a set of virtual machines that must be maintained at a certain level of security and timely patching.

The approach is as follows:

We select the repeat function for applications and centralize them in a cloud on the basis of the standard services – logging, database, messaging, etc. For such services, there should be separate systems that work, of course, in your virtual machines or public clouds.

Pools of virtual machines:

What is now implemented in the VMware View is that, we have a pool of deployed virtual machines that users can retrieve and return to the pool on demand (Check out – Check in). This approach allows access to the requested service or data from maximum speed users.

The factory of virtual machines:

Here we propose a standardized approach for  deploying services in virtual machines – something that VMware is now making vCloud Director and the Request Manager. The user should have a frontend, where it sets the basic required parameters to it, and let the rest be configured automatically or by the IT staff of the company (e.g. cloud services providers).

Of course, there should be an API for unification process between heterogeneous clouds.

Virtual machines that implement the PaaS (VM App). Why would you use someone else’s platform-as-service (PaaS), if you can deploy it yourself in virtual machines based on your IaaS (Infrastructure as a Service)? These are the machines for PaaS and proposes to deploy in the inner cloud. They have a very good scalability and can assist you in building the infrastructure on top of applications.

Virtual Service Machines:

Scaled cloud infrastructure using templates and VM is good, but also follow the concept of centralization of common services (e.g. directory service, or machines that provide storage resources). Here also, an opportunity to simply scale such virtual machines should be given, and they must have the highest availability, since they depend on other machines (workhorses).

The flow of services in virtual machines:

In the cloud infrastructure, where the modular nature of the applications is of great importance,the virtual environment provides much more flexibility than the physical. Therefore, it is good to build cloud services based on already existing services in order to not to carry out work on the deployment and configuration again, especially as templates and snapshots of virtual machines is easy.

Cloud broker:

If you are not able to trade on the exchange, then you hire a broker, who plays on your behalf (and at your expense), but if you plan to use the services of several service providers, you will need a software that can provide connectivity to third-party services and interaction with them in your “virtual” infrastructure (here the word virtual is in quotes because it is the second level of virtuality – that is the union of virtual resources from different providers). For example, a broker can do vMotion from one provider to another based on available resources, quality of service or even a rental value of virtual machines or applications.

By this I mean a service that will provide you information about your global cloud infrastructure. It will monitor vital functions of information systems to provide a comprehensive information about configurations and will be interacting with services on standardized interfaces.