Selecting the right database requires a careful study of your institution's needs, the complexity of your campus infrastructure, the resources and skill level of your IT staff, as well as the capital you have available to invest today and in the future. It is a multifaceted decision that is much more complex than simply matching the size of your school to a database. Today, the most popular database choices for higher education are the Microsoft SQL Server and the Oracle database. To make an informed choice between the two, institutions should consider four areas: reliability, scalability, security and total cost of ownership.
The first step is to evaluate the features that are available in the database to ensure the integrity of your data, and to ensure that your data is safe in the event of a system failure or disaster. Both the Oracle database and the SQL Server provide similar capabilities for data backup and recovery. The difference is that the SQL Server 2000 only runs on the Windows platform, hence, it is highly optimized and integrated with the operating system. This allows it to take advantage of features inherent in the operating system, such as failover clustering, with minimal effort. In contrast, the Oracle database employs Real Application Cluster (RAC) technology that enables all servers to share and process data from a single database. The applications recognize that there are multiple servers and automatically do load balancing across them. As a result, applications will continue to run even if a server goes down.
In general, a system's scalability is measured in both its ability to manage increasingly large volumes of data and increasingly large numbers of users/transactions without compromising the system's overall usability and manageability. The two approaches for dealing with scalability are scale-up and scale-out. Scale-up is implemented by increasing the computing power of a single machine. This can be achieved by adding CPUs, memory, or enhancing the networking components and storage systems. Scale-out is achieved by adding more servers to the system so that the workload is spread over a number of machines.
The SQL Server is able to scale-up on systems with up to 32 CPUs and up to 64 GB of memory. It can also automatically take optimum advantage of any available resource on the system without the need for manual configuration. The Oracle database is able to scale-up well beyond 32 CPUs and greater than 64 GB of memory, thus allowing it to achieve a higher net throughput. Scale-up is the preferred approach for most institutions because it offers the simplest implementation. Also, advancements in chipset and server component technologies can enable a single server to exceed the requirements of even the most demanding environments.
The vendors implement scale-out differently. Oracle's RAC technology allows institutions to increase their processing power by adding servers to an existing cluster that the existing database will recognize and utilize. The Oracle9i database scales on all hardware platforms; from single processor and mid-range multiprocessor systems to large-scale SMP, MPP, mainframe and clustered environments. The Microsoft SQL Server implements the "shared-nothing" architecture in its scale-out approach where data from a single database is divided into multiple segments, with each segment residing on its own server and storage unit. These servers and storage units are not related, and the configuration and operation of each unit is independent of the others within the same cluster. Data is presented and accessed in a unified format with the use of Distributed Partitioned Views (DPV) that manage the access, retrieval and insertion of data at the back-end, regardless of which server it physically resides in. …