Cloud computing discussions always start with the IPS Category: Infrastructure as a service (IaaS), platform as a service (PaaS), and Software as a service (SaaS). The advantage of this classification is that it is easier to understand and that the assessment requirements can be properly developed.
Want an application? A single purpose application (human resources, finance, printing, etc.) can be found in software as a service provider.
Want to write your own application? You can find an infrastructure, a service provider, that allows you to create your own custom applications.
While focusing on the application, want to understand how to use the software advantage of others to manage this concept? Then you can look at apps like Google App Engine that will give you a sense of PAAs.
These classifications are always added as a result of the lack of strong representational products, but the situation is changing.
Cloud providers in various fields are beginning to focus on future research and development battlefield--paas. They have come to realize that this is a crucial market – and that winning in this market will have a huge advantage. At the same time, this market also poses some major challenges to users.
This evidence is on our side. To date, Amazon has been the most successful IaaS provider. Although Amazon has not announced their services as "platforms", they have been enriching their core services with additional additional functionality that has a profound impact on the services provided by the company, helping to create applications more quickly and manage resource collections more easily. Think about the RDS for managing and expanding the database, the direct connection for securing external application access, the virtual private cloud for isolating applications in the AWS Data center, and the Cloudformation for application management.
At last week's Dreamforce conference, Salesforce listed their PAAs services. These projects are based on the Heroku companies they have recently acquired. Although these projects were once geared towards the Ruby on Rails framework, Heroku has begun to expand them to support Java. They are also integrated with Salesforce's database.com. With the integration of local data and Salesforce applications, it is supported by the database Rights option. Salesforce may call it a "social enterprise," but the entire service project is clearly designed to provide a common platform for application development.
Of course, not only are big companies targeting the market. Many recently established small companies are also offering a slightly different architecture for creating cloud applications. Although the companies claim they will use open source and cloudy models, people will continue to pay attention to their transmissions. In my opinion, cloud providers will adapt to every platform to prevent the portability of applications from being affected.
The impact of this approach is that simple IPs classifications will be quickly broken and subdivided into complex cloud computing worlds. In this world, each provider seeks a solution that can cover most of the customer's computing needs. Your SaaS provider wants to help you write your application. Your IaaS provider wants more functionality in its infrastructure to make your developers more productive. In this new cloud-computing world, this ambiguity will undoubtedly pose a challenge to companies, which are more difficult to understand thoroughly.
The challenges posed by PAAs
Why is it that PAAs is a major challenge for users? Simply because the strength and productivity of this platform has brought new problems to the enterprise. Companies may not be aware of these problems until they deploy a large number of applications.
These issues should be considered when IT managers evaluate their PAAs options:
1, lock. The integration of PAAs architectures with CSP architectures is far more difficult than installing applications into a virtual machine for a provider. If an application is internally dependent on the service provider's services, then extracting the application requires checking the code from the very root, rather than simply installing a packaging tool on another provider. The productivity value you get from the PAAs provider matches the level of lock-in you have with a particular service item. I don't think locking is purely negative. In my experience, companies are willing to be locked out because they offer more benefits. It is important to understand this when choosing PAAs, because PAAs will undoubtedly lead to greater lock-in.
2. Complexity. Each PAAs provider integrates their capabilities with their architecture, which is based on the provider's understanding of how the application should be designed. It is not important to determine how best to write and run your application in a PAAs environment. This is, of course, the biggest difference from the traditional local environment.
3. CSP (cloud service provider) difference. As mentioned above, a large number of PAAs architectures claim to provide an abstraction layer that hides the details of cloud providers from application developers. The possibility of applying abstract concepts to the actual work is placed aside to ignore the meta application software features, which can lock users into any project. Most of these functions are provided by the CSP and are focused on running instead of application editing. CSPs differentiate themselves from other providers by this level of functionality, and they lock you up at the operational level, not at the code level. Don't think this is going to happen. The first idea of a cloud provider is "How do I make myself different from other providers?" Because they are afraid to be computational power similar to the "dumb pipe."
4. New skills. Your application developers need to learn new framework knowledge and know how to develop and apply them. Although early cloud deployment companies have a large number of skilled developers who can quickly learn new skills, it is a human capital challenge for other companies to quickly acquire new skills.
5, guided by the new framework of the current practice. Most companies have identified the definition of architecture, methods, and operational practices. These must be evaluated and revised according to the new architecture. In fact, the problem already exists in the IaaS cloud service. As the new architecture leads to more and more guidance points, the situation will deteriorate further.
These questions may be seen as a reason to reject cloud computing. Not really. In fact, every new platform will have some big problems, whether it's microcomputers, PCs, cloud computing or laptops. It is important to recognize the pros and cons of each new platform and be prepared to face them. It is also important to remember the lessons and experiences that have occurred before. As the famous American writer George Santayana, "those who forget the past are doomed to repeat it."
(Responsible editor: Lu Guang)