Portability is important for Cloud Software

3 cloud for developers

There are fundamental changes of how a software developer writes code inside a cloud. In past, a developer had a software package with specific operating environment to target and had to have knowledge around that operating environment APIs. However SaaS developers at the time being prefer popular and promising development standards which are open and platform-independent due to the fact that they want to target numerous users and in addition, they want their SaaS to be easily portable and extendable e.g. Paypal Online Component, Amazon Store API or Gmail API. Another issue for SaaP developers was that developers were faced with limitations on their clients computing resources inside client infrastructure. Finally, developers had to round ideas around a product, however in SaaS they are developing a service. More clearly, after software was delivered to customer, there were not much interaction between customers and developers unless in case of a bug fix or a new release which meant a new product in the sense of SaaP. While developing SaaS, customers frequently need improvement on their services and as long as they have a contract, this interaction exists. In service-based development, since software is being utilized by vast number

of users, therefore scalability is another issue for SaaS hosts.

3.1 Agile Development

One of most critical part of a SaaS project is the choice of development methodologies. Based on the unique features of a SaaS which was mentioned in section 1.2, agile methods are embraced inside SaaS projects. Reasons are that Agile methods are divided into a set of tasks and each task constitutes of several iterations. Each task consists of a set of goals and is completed within a number of iterations. More specifically, the benefit of using an agile method is that after the software designer will agree upon the task design, and the developers and tester start coding and testing respectively, since tasks can have more and more iteration, there can be additional features and perspectives in each cycle that can be added to the task over the time. The result of applying such process shall be that the changes are no longer infrequent, easy and financially sufficient to make. There exist numerous variants of Agile methodologies, some of which are XP, Scrum, Feature Driven Development and Lean Software Development.

CS Scalability, CS Portability

3.2 Portability and Scalability in SaaS

As mentioned in section 1.2, due to the scalability factor of a SaaS project, developers tend to utilize platform-independent frameworks in order to increase portability and cost efficient scalability inside their SaaS applications. As an example there will be developers preferring to work with Objective-C inside Apple’s iCloud, however since Objective-C is coupled to Apple’s products one cannot benefit from using it outside iCloud. As another example, Microsoft ASP.NET can also be a wise choice while working with Microsoft Azure service, however in case a company receives a better offer from deploying his service on Google App Engine, developers should re-writes at least some parts of application in Java or Python, if not all. For scalability, development team has two options available: Vertical and Horizontal Scalability. Vertical Scalability is the idea of enabling severs with CPU and memory power in order to handle a high load of requests. On the other hand, Horizontal Scalability is to have multiple unit of works and enabling them to handle loads of request, as if there are working as a single unit of work. Vertical Scalability is expensive however is an easy choice. On the other hand, Horizontal Scalability is financially cheap however requries some considerations. Considerations are on how all layers such as CPU (i.e. processing power), Storage layer (e.g. Clustered file systems or Amazon S3), the database layer, Application Layer(e.g. Tomcat clustering), the Web Layer, Load Balancer , Firewall can scale out synchronously. For more information on Scalability and Portability, see [2] .

Cloud Portability

References:

[1] Martin Fowler. Patterns of Enterprise Application Architecture. Addison- Wesley Professional, 2002.

[2] Armando Fox and David Patterson. Engineering Long-Lasting Software: An Agile Approach Using SaaS and Cloud Computing. Strawberry Canyon LLC, 2012.

[3] David S. Janzen and Hossein Saedian. Does test driven development really improve software quality. IEEE Software, 8(3):77–84, 2008.

[4]: Praising Gaw's blog, What is the difference between cloud computing and SaaS, http://cloudcomputing.sys-con.com/node/612033

[5]: STEVB's blog, Differences between SaaS and cloud software, http://www.mindtouch.com/blog/2008/05/28/differences-between-saas-and-cloud-software/

Interested in reading about Android OS Securitiy, check out the following posts:

Android Attacks in a Nutshell
  

Cloud Computing: Developer's View (Part III)

2 software development process

During 1980-2000, the industry was focusing on traditional development process methodologies such as Waterfall and Spiral methods. In these approaches, Rapid Application Development excluded, a stable version needed to be released firstly as alpha and beta versions. After that and in order to have the SaaP bugs fixed, customers had to either play around with the software or utilize it inside a production period. This was a long and risky process from customers’ point of view, since some bugs could have unrepeatable damages to data and Operating Systems beside the software itself. The latter was due to the fact that in traditional software development methodologies, verification phase and quality assurance was mainly being done after implementation phase. While there were some successful cases of software products with traditional methodologies, the following was a list of major issues in developing based on them.

• Before developing the software, designers needed to come up with a detailed full design plan. The issue was that some requirements were raised as the development process was evolving and could not be forecasted in the earlier design phase of the project.¹

• Beta releases needed months to become stable versions and attract all customers without worrying them.

• Changes in software were costly and infrequent, therefore customers could not have the set of features they needed in short period of time. Due to criticism of developer communities and many examples of software running over time and budget, this approach was nearly abandoned over time. In 2001, a group of developers aimed to come up with a lighter-touch lifecycle. Agile Alliance stated the following in the Agile Manifesto:

We are uncovering better ways of developing software by doing it and helping others do it. Through this work we have come to value:

• Individuals and interactions over processes and tools

• Working software over comprehensive documentation

• Customer collaboration over contract negotiation

• Responding to change over following a plan that is, while there is value in the items on the right, we value the items on the left more.

While working on a SaaS project, developers will be asked to change the code frequently. Customers signing a contract for a SaaS want constant improvement of their service. Frequently developers are greeted with new requirement which changes components of the project. Also, in case a SaaS developer develops a perfectly architectured and coded application, it shall not be accepted by the customers without his verification. That is the reason that in Agile methodologies, verification is separated from validation. Since validation is how to architect and code a software correctly, verification is the question of following scenarios as customer needs their service to be. It should be noticed that a service is coupled to a domain, bringing so much emphasis on agile verification methods.

1 Steve McConnell in his book ’Code Complete” refers to design as a ”wicked problem”, i.e. a

problem whose requirements and limitations cannot be entirely known before completion.

Cloud Computing Models: SaaS versus PaaS

1 SaaS vs. SaaP

1.1 History

Traditional SaaP developers was focusing on how to access Operating Systems (OS) APIs and how to handle those APIs in order to utilzie them inside a software via the development framework in use. After a while, the question of application portability came under discussion in early 90’s due to numerous

operating system segmenting the OS market. In 1995, James Gosling introduced the platform-independent Java and became the first scientist to tackle portability issues. Java programms were running on a virtual machine enabling applications to be ported in all Operating Systems. After that more

and more frameworks started to follow the path of Java to enable platform independency such as .NET framework by Microsoft, different implementation of Python and etc.

With Internet gaining huge popularity and profitability in the Dot-Com Bubble era during 1995-2000, IT industry started to invest remarkably in Web Infrastructure. However there was no disucussion about SaaS yet. A similar concept called Utility Computing was already introduced by John McCarthy in MIT in 1981 as ”an imagination of nearly infinite amount of computational resource”, however the SaaS dialect evolved after the release of Gmail, which was a successful implementation of SaaS. What was unique  about Gmail was the ability for the user to extend its API and utilize it as an email service inside his own applications. After Gmail gained popularity, companies started to invest more and more in SaaS engineering. In 2003, Amazon with its EC2 service truely started the Cloud industry. Not only Amazon  offered Infrastructure but also platforms to be used as a service. This initialized the cloud dialect after their service models were introduced, i.e. Platform-as-a-Service (PaaS) and Infrastructure-as-a-Service (IaaS). Recently, Microsoft has launched its Office 365 SaaS project, in order to allow users to pay for their use of software, instead of paying the whole license at the purchase time. This had made software development process to evolve in a complete different path than it used to be.

1.2 Features

While developing SaaP, vendor has an approximately large set of concerns in most cases. First is the application requirement regarding operating environment, hardware and software dependencies. Second is the application portability issue. Each framework has its own boundaries while dealing with system APIs and therefore either team members should be assigned to test the application against different operating environments, and re-write the software components or limit the user to specific functionalities in case of incompatibility. As a result, due to tight-coupled software and hardware package, changes in software are costly and are avoided in case of appearance of bugs that have low-severity level of destruction toward the system or customer dissatisfaction. Last concern of vendor is the customer’s data. Whether sensitive or not, customer’s data could be damaged by misuse or fatal bugs that SaaP-based applications could create.

With players involving more and more in the cloud market, it is apparent now that SaaS experience is completely different experience than it was in SaaP, either reviewed from a customer or a developer’s point of view. Some of the highlights of SaaS unique features can be listed as follows:

• There is no need for installation of any software rather than a standard Web Browser.

• Customers are working with the latest release of software. They may not even know this fact themselves, therefore they cannot choose not to.

• Customers do no longer have to worry about backing up their data as it is now the responsibility of SaaS providers to operate data back-up processes.

• SaaS software runs on a uniform and tightly coupled hardware, chosen by developer and therefore compatibility is no longer an issue. It should be remembered that no longer binaries are distributed amongst operating systems as it did in SaaP.

• Developers can upgrade the software and underlying hardware as frequent as they need to, without changing the API of the application.

Cloud Computing: Developer's View (Part I)

0. Introduction

Every IT guy has already been taken a glimpse at cloud computing and its service models. Apart from the fact that the cloud computing idea initialized from SaaS, IaaS, PaaS, nowadays the concept of Everything-as-a-service sounds interesting (if not promising). 

After the initialization phase of an idea or a technologies, each industry moves forward to its member's role definition. While network administrator, hardware engineers has already gotten their role to start developing something on the cloud, one group has nearly left out: Software Developers.

It is obviously negotiable that Software Developers will now mainly work as SaaS engineers but there is one single issue: different elements between a SaaS and a web application is relatively negligible in most cases. It sounds fairly logical that nearly every subscription-based web application has the potential to be deployed as a SaaS, meaning that software-vise, nothing really new is happening. More precisely, with having sociability potentials aside, SaaS engineering is actually reusing software pasterns which are already been used in SOA and web development alongside agile programming models such as XP. 

Figure 1. Scale-able SaaS elements [2]

After all that is being said, there comes an idea that what if the potentials of cloud computing are yet to be realized from software developer's point of view. The terminology "Cloud Software" is now becoming more and more popular amongst IT professionals. There is a discussion that Cloud Software is not basically the same as SaaS: they are either massively scalable applications [4], or the type of application in which user has full control over his data replication or the application is running into or off to organizations own cloud infrastructure [5]. While first definition is not understandable clearly from its definition,  latter looks more like to SaaS running on a private or (at least) a hybrid cloud.

To learn more about cloud computing, continue with:

[1]: Cloud Computing Models: SaaS versus PaaS

[2]: Cloud Software Development Process

[3]: Cloud Software (CS) Portability and Scalability