Queries From Hell
A blog about data warehousing, correlation databases, associative and incremental queries, value-based storage, metadata, on-the-fly indexing, automatic data-driven schemas, BI tools, data mining, visual mapping, pattern recognition, and the limitations of standard SQL in answering "queries from Hell."
Or, how to discover what you don't know you don't know.

Database in the cloud - blue sky?

Posted at 10/06/2008 11:10:00 AM
At the VMworld meeting in Las Vegas a couple of weeks ago, virtualization of everything seemed like the norm. Cloud computing was the latest greatest thing and is seemed like the entire world was lining up to move into the cloud. While there are many clear benefits to cloud computing, there were discussions about databases in the cloud that left me wondering if people really thought this through. Certainly XML files containing datasets used in cloud computing functions are not only valuable, they are required for almost all cloud computing applications.

But is there a place for real enterprise data stores in the cloud? Let’s first look at why you would want to put anything in the cloud. Most of the reasons that I have heard relate to ease of managing distributed applications. Ramping an application up from 50 servers to 350 servers in just a few days is possible in the cloud and impossible in the computer room.

Clearly, one of the important benefits is the accessibility of infrastructure in real time. Infrastructure, though, is a very broad term. It might mean access to electrical power, communication lines and brick and mortar facilities for housing servers. But it can also mean access to enterprise data stores.

The question is does access to enterprise data stores in the cloud imply that the database is in the cloud or access to a managed central database from cloud based applications good enough? Why would you want an enterprise database in the cloud?

One reason might be to provide support for wildly variable demands for the computing power needed to meet database processing requirements. However, managing a database in the cloud would seem to be impossible, especially with wildly varying loads. How could storage be optimized and how could partitioning be managed? It would seem that performance could never be tuned and a database in the cloud would provide universally poor performance.

There are many other issues that would need to be addressed, like access control and data quality, but the DBA’s job would seem to be the most difficult obstacle to overcome.

Add to Technorati Favorites      | Blogger Comments (0) |