Cloud Connect Conference Leads Cloud Development

One of the earliest cloud computing inaugural meetings was the Cloud Connect conference in Santa Clara, California. To date, one of the more contentious issues of the conference is "Do not use relational databases to preserve the persistence of data." Known as the "NoSQL" movement, its purpose is to use other forms of databases to process large-scale application data more efficiently. However, I have already published a number of articles on many "large-scale data" that appear in cloud computing. But the sport is more special and it will be a way to push the data back to simpler, and perhaps more physically effective, models for physical storage.

NoSQL systems typically store data in memory at run time or read data from many disks in parallel. One problem is that the "traditional" relational database does not provide this pattern and therefore does not provide the same performance. In the past that database, if only a few GB of data, the problem is not obvious, but many cloud computing database has more than 1TB, there will be more large-scale database will be used to support the ever-evolving Cloud computing system. Working with large-scale data on a relational database is a taboo for the military because SQL requests consume a large amount of CPU cycles while processing data and can result in a large number of disk reads and writes.

If you feel like I've heard it before, then I tell you that you're right. As early as the 90s of last century, some progress has been made in the object database and the XML database. Although many non-relational databases can indeed provide better performance then, many enterprises hold the key to relational databases such as Oracle, Sybase And Informix. However, due to the high cost and risk of migrating from relational databases, and the relatively small size of the data, relational databases are virtually unrivaled.

However, cloud computing has changed everything. The need to process large amounts of data in cloud computing has led to the application of new database processing methods to the older models. MapReduce is the basic way for Hadoop to process data based on the "share-nothing" database processing model a few years ago, but now we have the processing power, disk space, and bandwidth to implement it.

I estimate that the development of cloud computing will reduce the use of relational databases. This is not new, but this time we really need to change.