Comparison of Big Data revolution with gold fever is already pretty crammed, but apparently we cannot escape from it. It too accurately describes the situation, including the diversity of problems and choices faced by “miners”. Platform choice, choice of provider, selecting the type of integration … And all of this is overshadowed by thoughts about the prospect of such a transformation. Figures speak themselves, but when the market situation changes radically in a month, flexibility becomes a major priority.
Depending on the manufacturer
There are different ways to reduce risk. First, it should initially raise issues of data portability even in the analysis of vendors and contract options. That is, if you decide to roll your presence on one platform and switch to another, then how much will the data transfer and conversion process cost. If you do not know, you can be in a situation such as “big investment”, and to use a certain product / platform is necessary, since all of your transactions will be built on it and refusing to be extremely painful. Theoretical and technical cooperation between the platforms is not a big problem from a technical standpoint. Establish a process of data so that they are stored on a single platform, processed and output to another via a third party application is in principle possible, but not profitable for vendors who prefer to provide comprehensive solutions “and from” that it is easier and more profitable for them as well as ties to customers cooperate with them. And as soon as the migration cost is too high, they will try to transfer customers to more expensive service packages. Let’s try to see the situation from the point of view of the buyer …
Levels of abstraction
To a large extent, dependent on the partner defined service model, which is built on cooperation. In general – the higher the level of abstraction from the hardware and less effort is required on the client side, the stronger the lock-in. Consider the examples:
SaaS Model – (example: MS Office 365) using such tools for business purposes, you should be fully aware that they are provided “voluntarily” developer. Make any modifications to the software alone is not possible, and in the case of withdrawal of support software you will have to look for another service. In this case, you will probably be able to export the data, but perhaps they will be in an unsupported format and most likely will not be able to automate the process (again with the caveat – without the participation and support of the company, the platform holder).
To move, the flow will stop. Acceptable level of safety for the larger business processes. Scaling is easily done within the established framework, the excess is either impossible or by individual appointment with the vendor. Minimal integration costs. Number of small vendors.
PaaS Model – (Google App Engine, Windows Azure, Amazon, ESDS eNlight Cloud Hosting) client greater freedom in the use of the platform, but the possibilities are limited by the tools that are available to developers. Nevertheless, the ability to adapt to the specific requirements considerably wider. The export data provided depend on the application developer tools. Automating the process is possible. Level of security sufficient for most operations. Scaling limits depend on the agreement with the holder of the platform. Moderate integration costs. Average number of vendors.
IaaS Model – (Amazon EC2, Google Compute Engine, Rackspace) client has at its disposal a virtual infrastructure, which can perform any task and its interesting applications. Ability to export data ample chance failover, connect / disconnect power without interrupting work. Ability to fully automate most processes. Extensive customization capabilities. Level of security in taking appropriate action is not inferior to that in the work on their own equipment. Scalability is virtually unlimited. Significant integration costs. Maximum number of vendors, strong competition between them. As seen from this “analysis”, the lower the level of abstraction, the less dependence. This is explained as greater harmonization platforms IaaS, and the high cost of developing SaaS-complexes, why a relatively small number of vendors offering such solutions and they are comparatively less than competing with each other.
Does this mean that it is necessary to seek opportunities to transition to IaaS? Yes. No. Possible. IaaS model is better suited for large businesses that can afford to (and have the need to) develop their own applications. For small and medium businesses are well suited model PaaS and SaaS, which provide an acceptable user experience and cost reduction. Later among the three galleries can wedge the other players, which provide unique opportunities. For example now, in the field of cloud storage, private cloud projects (ESDS Private Cloud) successfully compete with proprietary solutions. Maybe after a while we can expect similar progress in the field of services of general purpose – such as online office suites.