Database Efficiency{1}


by Mike Y

           Cloud computing is becoming increasingly popular. Platforms as a service (PaaS) can rent resources from infrastructure-as-a-service (IaaS) to have level of service required without investing too much or too little. The article concentrates on multitenant databases where a service will try to reduce costs by grouping tenants. This can be achieved by using SmartSLA to efficiently manage data. The research team evaluates SmartSLA by comparing it to the TPC-W benchmark. The tests show that SmartSLA “can provide intelligent service differentiation according to factors such as variable workloads, SLA (service level agreement) levels, resource costs, and deliver improved profit margins.”

           Databases must be managed with limited resources to keep costs low while keeping services fast all while maintaining the integrity of the data. This is important today because an increasing amount of information is stored online. Instead of keeping photos locally, many people will upload it to sites such as Facebook or Flickr where it can be shared and accessed from anywhere. If the data is not managed efficiently, the photos can be lost or inaccessible and will lead to an unprofitable business (and enraged customers). In a world where a site can have millions of connections a day, even the slightest increase in efficiency will quickly add up.

           There are services that rely on cloud computing such as Google’s services (calendar, documents, photos etc.), Google’s Cr-48 laptop, and OnLive gaming service. As cloud computing increases, so will the amount of data that will have to be transferred. I believe intermittent or slow connectivity and data loss will be eventually be completely unacceptable as technology advances and its use becomes more widespread. In addition to being fast and reliable, databases must also be secure so that private information is not compromised. Another important factor to consider is if the system will scale well with an increasing amount of users.

Source:
Xiong, P. (2011). Performance prediction for concurrent database workloads. Intelligent Management of Virtualized Resources for Database Systems in Cloud Environment, 11(16), 87-98. doi: 10.1109/ICDE.2011.5767928