Extreme input and output techniques refer to techniques of managing and controlling inputs and output scales in data processing and other processing units. This regards implementation of parallel applications that help reduce risks involved with events tracing on extreme scale inputs and outputs (Smallwood, 2005). To manage inputs and outputs it’s rational to define means of capturing large volumes of data, processing of the data and creating innovative outputs.
Capturing detailed information otherwise called – big data is an arising requirement due to expansion in productivity growth, research, consumer surplus and innovation. There are various sectors which are key sources of big data which include; health sector, manufacturing industries, astronomy and genetic data. These fields require capturing of the large volumes of data involved. Big data require techniques that effectively handle large volumes of data. These techniques are; crowd sourcing, massively parallel processing (MMP) databases, distributed file systems and databases, and cloud computing architectures. The highlighted methods enable capturing of massive volumes of data that is implemented through these strategies; performance management, data exploration, social analytics and decision science (Smallwood, 2005).
Performance management involves understanding big data where multiple databases are integrated, and database tools designed to help users make use of the large volumes of the data. Data exploration regards the predictive modelling techniques on the big data. This incorporates the use of statistical and analytical techniques on the big data captured to make effective decisions on business processes and planning. According to the 1963 proceedings on input-output techniques, focus was laid on collection of data and making quality use of the data through controlled data exploration techniques. Social analytics defines social metrics on the big data e.g. influencers to business processes. This separates metric data (transactional data) from other data. This metric data is used to measure the impacts of processes. This leads to the decision science which involves experiments on non-transactional data where major decisions on the big data are made (Huang, 2009).
The captured data needs storage and processing before it gives any outputs. The data is stored in integrated and distributed file systems and databases. From the database, the data is sourced via network connected terminals from where users access the data. MMP (massively parallel processing) techniques are used to process the data to give the requested outputs defined through query mechanisms on the data. MMP involves segmentation of data into parallel sets that utilise permutations and complex interrelationships to process data independently in segments. Different queries are designed to source various aspects of the data making the data sensible and manageable to use. This data require minimum storage devices equivalent to 1000 terabytes to hold the large data volumes and minimum memory of 4 gigabytes to process the data fast and efficiently.
Works cited:
Brody, A, and Anne P. Carter. Input-output Techniques: Proceedings. Amsterdam: North-Holland Publishing Co, 2008. Print.
Smallwood, David O. Extreme Inputs/outputs for Multiple Input Multiple Output Linear Systems. Washington, D.C: United States. Dept. of Energy, 2005. Internet resource.
Huang, De-Shuang. Emerging Intelligent Computing Technology and Applications: 5th International Conference on Intelligent Computing, Icic 2009, Ulsan, South Korea, September 16-19, 2009 : Proceedings. Berlin: Springer, 2009. Print.
