Recently, I came across a nice blog on why Amazon thinks big data was made for the cloud. It talks of how big data and cloud computing will work hand in hand to create a central platform for communities to share huge Data Set. In Memory Data management such as HANA enables this symbiotic relationship between the cloud and big data by facilitating on the fly reorganization of data
Cloud has multi-tenant data. Multi-tenant data are managed primarily by 3 different approaches:
- Separate Database for customers
- Shared Database, Separate Schemas for customers
- Shared Database, Shared Schema for customers
Of the three approaches, the shared schema approach has the lowest costs, because it serves largest number of tenants per database server. Also, the administrative and hardware/software costs are drastically reduced. But, it comes with one complexity: As the customer isolation is the minimal, stringent database management is required to ensure that tenants can never access other tenants’ data, even in the event of unexpected bugs or attacks. Dynamic reorganization of Data is one of the prime requirements.
In Memory Databases such as HANA leverages the positives of columnar databases so that
- New Attributes can be added easily vis-à-vis a row based database architecture
- Locking for changing the data layout in only required for a short period contrary to row based architecture where the entire database or table would be completely locked to process data definition operations
I feel that there is a great applicability of In Memory Data Management in cloud. Do give in your comments/ feedback on this piece.