Big Data Advantages And Disadvantages

Data is invaluable in product prediction, scientific exploration, business intelligence, and so on. However, the sheer quantity and velocity of Big Data have caused Problems in data capturing, storing, maintenance, analysis, search and visualization. The management of a huge amount of data is particularly challenging to the design of database architectures. In addition, in many instances when dealing with Big Data, speed is not an option but a must. For example, Facebook makes an average of 130 internal requests sequentially for generating the HTML for a page [1], thus making long- latency data access unacceptable. Supporting ultra-low latency service is therefore a requirement. Effective and smart decision making is enable with the utilization…show more content…
Recent advance in advances in hardware technology re-kindled the interest in implementing in-memory databases as a means to provide faster data accesses and real-time analytics [14], [15], [16], [17]. Most commercial database vendors (e.g., SAP, Microsoft, Oracle) have begun to introduce in-memory database to support large-scale applications completely in memory [18], [19], [20], [21]. Nevertheless, in-memory data management is still at its infancy with many challenges [22], [23], [24], [25], and s completely in-memory design is not only still prohibitively expensive, nut also unnecessary. Instead, it is important to have a mechanism for in-memory database that utilize both memory and disks effectively. It is similar to the traditional caching process, which is however the other way around: instead of fetching data that is needed from disk into main memory, cold data is evicted to disk, and fetched again only when needed. In this case, main memory is treated as the main storage. While

disk acts as a backup. We call it as “Anti-Caching” to emphasize the opposite direction of data movement. To goal of “Anti-Caching” is to enable in-memory databases to have the “capacity as data, speed as memory and price as disk” [27], [28], being a hybrid and alternative between strictly memory-based and disk-based databases.

1.2. Contributions and Paper…show more content…
We give a discussion about the differences between caching and “Anti-Caching” in Section II. In Section III, we present state-of-the-art “Anti-Caching” approaches, and their basic mechanisms. We conduct a thorough experimental analysis of these approaches, with regards to the performance effect on the systems in Section IV. Section V summarizes what we derive from our experimental analysis, and presents some insights and tradeoffs on designing an “Anti-Caching” approach. Based on our findings, we sketch an easy-to-use general approach that can be incorporated into most in-memory databases easily in Section VI. Section VII concludes the