Data Management Strategies in Multiple Clouds By Aditya Abeysinghe
The “cloud” has been one of the most trending paths in systems and application deployment during the past few years. The cloud has built-in functions for computation, analysis and networking and it dominates almost all application-specific functions we use at present. With the growing use of cloud services, several types of clouds have been proposed. Not only single clouds, multiple clouds have also been researched. In this article, several multiple cloud types are explained from a data and application management angle.
Public and Private clouds
Public clouds refers to cloud services provided by a cloud service provider where computing services such as storage are shared with other users. With public clouds, multiple users may access the same resource at the same time. For example, multiple users may store data in the same file system in the same storage allocated. Therefore, privacy is low and the vulnerability for attacks is higher in public clouds. However, public clouds are often cheaper, easy to maintain and the provider often has several built-in functions for services they provide.
Private clouds are cloud services where services are provided only to the user and not shared with other cloud users. Private clouds can be hosted on a cloud service provider or hosted as a cloud on-premise. When hosted on the service provider, all resources are dedicated to the user. Therefore, privacy is higher and the vulnerability to attack is lower. However, maintenance needs to be done by the user at most times and due to this, the cost of private clouds are higher.
Hybrid clouds and Multi clouds
Due to advantages and disadvantages of both public and private clouds, use of multiple clouds has been a trend in cloud-based services. Hybrid cloud, multicloud, and intercloud are three options for multiple clouds.
In a hybrid cloud, a public cloud and a private cloud are used. Therefore, it is a mix of the pros and cons of both types. However, hybrid clouds are costly for small to medium businesses and are often the norm for large scale businesses. Also, hybrid clouds are frequently not opted for businesses whose data are protected over regulations and privacy such as financial and governmental institutions.
In a multicloud, multiple public clouds are merged to provide services. However, it is also common to use a private cloud with public clouds. With multiclouds, the same application is hosted on multiple service providers. For example, the data analytics platform of a business may run on one public cloud of one cloud service provider and a private cloud of another service provider. The difference between hybrid and multiclouds is that in hybrid clouds, components of an application is hosted on public and private clouds. Integration between these components is necessary to function the entire application. However, in multiclouds, the same application is deployed on several public or private clouds.
Interclouds
With the intercloud, services of the same application are hosted in different clouds. For example, the data analytics module of an application is hosted on one cloud service provider, the interface of the application is hosted on another cloud service provider etc. Data are transferred between these services to provide the complete functionality. With intercloud, vendor lock-in is reduced and often at least one component can provide the functionalities even if other components fail.
Evaluation of different multiple clouds
Performance, disaster recovery, security and monitoring are some factors that need to be considered when choosing multiple clouds for hosting/deploying applications. There is hardly a best multiple cloud option. Multiple options need to be evaluated based on their functions and the type best matching the needs should be chosen.
Disaster recovery is an important aspect of clouds as data need to be recovered and made available in case of failures. In this aspect, multiclouds are the best option as the same application is deployed in multiple providers and no integrations between clouds are required.
Data security is another area that is important in cloud environments. Hybrid clouds are the best option when considering security as sensitive data can be protected within the private cloud with less cost compared to that with the intercloud approach. In intercloud, additional cost is required as components are distributed between clouds and monitoring and integration between clouds is required.
Interclouds are still a theoretical cloud type which are hardly adopted in consumer clouds. However, they can guard against failures as failure of a component is rarely affected to other components due to use of multiple vendors.
Image Courtesy: www.testqtech.com
