The year 2009 saw an increase in cloud services adoption and accelerated growth is expected this year. A survey conducted by Web-based service provider ReliaCloud indicated 83 percent of top U.S. firms said that they are already using cloud computing services, or that they have plans to do so within the next year.
The relatively high adoption rates of cloud-based services are rooted in clear business and IT benefits derived from migrating services and applications to the cloud. Benefits include CAPEX and OPEX (News - Alert) reduction, faster time-to-market of new services, infrastructure standardization and disaster recovery assurance.
The benefits gained by migrating services and applications to the cloud greatly depend on the type of cloud service an enterprise is planning to use. Enterprises today use a variety of cloud-based services, including cloud software services for CRM, supply chain management, development services, network infrastructure services, and cloud backup services.
In parallel to the trend of migrating to the cloud, some enterprises have started the process of deploying private clouds. This enables IT managers to maintain control over their network infrastructures while benefiting from a completely virtualized IT network infrastructure.
For cloud service providers or private clouds to accommodate the growing number of new customers or hosted applications, they must be able to quickly adapt to the growth of new users and traffic resulting from an increased user base. This is achieved by creating an elastic data center virtualization infrastructure which can dynamically grow when additional computing resources are needed, or scale down when throughput levels drop and computing resources can be removed. The dynamic allocation of resources should be managed and monitored such that the application SLA will be guaranteed and transactions will continue uninterrupted.
To effectively manage an elastic virtual infrastructure, the IT managers of cloud providers should implement an Application Delivery solution which facilitates the automatic adding and removing of computing resources on demand, when application SLAs are below adequate levels. This application delivery solution should consist of two mechanisms – dynamic resource allocation and application traffic redirection.
The dynamic resource allocation mechanism continuously monitors the performance levels of an application as well as network resource availability and other business KPIs (Key Performance Indicators). Additionally, it also automatically adds/removes computing resources according to measured KPIs.
The application traffic redirection mechanism is complementary. As resources go up and down in the data center, distributing application transactions in between the available computing resources is key to successful application delivery. Therefore, the dynamic resource allocation mechanism should synchronize with the application traffic redirection mechanism, in order to redirect users and traffic to the newly added computing resources.
Cloud providers require optimal availability and responsiveness of their Cloud infrastructure, and therefore, will build multiple data centers in multiple geo-locations. The application delivery solution should be able to not only manage the computing resources in each of the data centers in which the application resides, but also be able to redirect users to the remote data center if the local data center has reached maximum capacity. Once the application performance levels in the local data center have reached adequate levels, the application delivery solution should remove the remotely added computing resources and update the application delivery logic to use the resources of the local data center.
Additionally, when the cloud serves users from distributed locations around the Internet, the quality of experience of users can vary considerably depending upon the location of the data center that serves a specific user. The application delivery solution should identify the location of the user and then redirect the user to the data center that provides the best SLA for that specific user. Optimization of SLA is the most important streaming applications and other latency sensitive applications.
To summarize, a truly application-aware application delivery solution will ensure that performance degradation is handled automatically in zero time and will facilitate OPEX reduction by freeing IT resources. Furthermore, such a solution will also take full consideration of available resources, current traffic loads and users proximity to data centers when redirecting traffic in the cloud. This will result in improved QoE for any cloud customer accessing the application.
Amir Peles is Chief Technology Officer at Radware