| Cloud Computing is becoming a mainstream | | | | other complex view of data was modeled in these |
| technology with almost all major industry heavyweights | | | | forms and the resulting performance hit was handled |
| backing it. As details emerge and the platforms and | | | | by de-normalization and other techniques. However, |
| technologies are refined, many software vendors are | | | | with the explosion of data that we have seen in the |
| evaluating the move to cloud. Being a fairly new and | | | | last few years, these models seem to be breaking at |
| emerging technology, there is a need to evaluate the | | | | the seams. |
| value of migration to cloud at minimal cost. | | | | So, the erstwhile taxonomy of entity, relationships, joins |
| In this article, we will look at two pieces of software | | | | etc. is being fast replaced by priorities like replication, |
| that a vendor can move to the cloud without a | | | | redundancy, partitioning and folksonomy. As you can |
| negative impact, and in fact benefiting from inherent | | | | imagine, these requirements actually end up requiring a |
| strengths of the cloud platforms. These are meaningful | | | | whole lot of computing power and scalability. Enter |
| and progressive steps towards the cloud that have a | | | | cloud computing. All cloud computing platforms sport |
| very good chance of being in the mainstream cloud | | | | multiple data modeling architectures. Amazon uses |
| when the air clears around it. | | | | SimpleDB, Google has BigTable and Microsoft has |
| Identify your Whole Product pieces and move them to | | | | SQL Data Services, which supports Queue, |
| a cloud-enabled data center | | | | (Non-relational) Tables and Blobs. Any part of your |
| Every product has bells and whistles that create a | | | | application that you think has been force-fitted into an |
| complete experience for the users. In product | | | | RDBMS structure is best moved to this newer |
| management parlance, this is called the Whole Product | | | | paradigm of data management. Some examples of |
| or Augmented Product, as opposed to just the | | | | such complex information are: |
| software. Following are some examples of whole | | | | |
| product: | | | | 1. Tagging: When I first started using GMail, I was blown |
| | | | away by their email "Labeling" feature. Labels allow |
| 1. All the rich content around a trip planning website | | | | you to "tag" your email by different names. Labeling is |
| such as city guides, attractions and images | | | | clearly superior to using folders, since if you tag an |
| 2. User Generated Content (UGC): The traditional | | | | email as Work, Project and Web 2.0, you have not |
| definition of Whole Product included only the stuff that | | | | one, but three different ways of reaching the email. So, |
| product managers and companies included to | | | | few years of GMail use later, I struggle today with the |
| complement the user experience of their product. For | | | | idea of having "folders" in my email client. Tagging or |
| example, the iTunes Store completed the iPod user | | | | labeling of information is just not practically suited for |
| experience. In the Web 2.0 realm, user generated | | | | RDBMS or hierarchical databases. So, if you have or |
| content is one such item. Moving out the augmented | | | | plan to use tagging or folksonomy in your product, you |
| product pieces to the cloud is a smart idea for two | | | | would do well to move it to the cloud. |
| reasons. Firstly, it is usually safe to say that the | | | | 2. Data-centric applications that have multiple |
| augmented pieces are an essential, but non-core part | | | | interconnections between nodes: This is quite |
| of your product that you would want to manage easily | | | | self-explanatory, but there are many data intensive |
| and efficiently without losing your focus on the core. | | | | applications out there that can benefit from a rethink in |
| Secondly, UGC and other content are a big part of | | | | the way their data is modeled. This harks back to the |
| user experience these days, and are inherently very | | | | force-fitting I alluded to earlier, and you would be clearly |
| elastic and unpredictable in growth. Cloud computing will | | | | making the right move by moving it to the cloud. Seen |
| enable you to manage and scale them very | | | | from the perspective of data modeling then, cloud |
| effectively. | | | | computing becomes a piece of an extremely |
| Move data with complex structural requirements to the | | | | important puzzle. You need scalability in using some of |
| cloud | | | | these newer data models, and cloud computing is your |
| Traditional way of looking at data has been in a | | | | best bet to be able to do that properly. |
| relational model or maybe a hierarchical model. Any | | | | |