5 thoughts on “Modern Databases of the 21st Century”
Leave a Reply
Modern Databases of the 21st Century
Select your location
English | EN
English | EN
5 thoughts on “Modern Databases of the 21st Century”
Leave a Reply
I totally agree that a lot of new database technology is entering the market – I’d say Hadoop is the most well known example of the new entrants though not the first.
But I see far too many discarding the relevance of the RDBMS in favor of one or more of the new technologies.
But some RDBMS are a lot better at handling various workloads than the rumor claim. An often overlooked example, that handles most of the requirements perhaps except storing “unstructured” data, is Teradata RDBMS with its massively parallel, shared nothing architecture allows for a mix of row-based and columnar storage. It also uses data temperature to automatically place data where it makes most sense – Hot data in memory, warm data on SSD-drives, tempered data on the fastest (sectors of) disks and cold data on the slowest (sectors of) disks. Due to the architecture you get linear scalability up to 61 PB of uncompressed user data.
Nice & thorough, thanks!
Thanks Ashish for an excellent quick summary of databases. Summarizing your article leaves below points:
1) Star Schema is used in OLAP systems, however an improvement can be made using column-column approach rather than traditional row-row approach. HP’s Vertica, SAP’s HANA and Amazon’s Redshift are examples.
2) Rather than having a database that does hard disk storage-memory writes and reads , thus wasting lot of time in that , one can opt for main memory and single threaded databases like Volt DB.
3) No SQL databases provides all together a different dimension of scalability, felxibility and supports various kinds of data models such as Cassandra, MongoDB and CouchDB.
Questions:
1. Can columnar databases be used in OLTP kind of systems ? If not ( not better suited) why ? If yes how ?
2. Main memory loading and single threading looks more like an charesteristic of a system ( or system architecture and RAM availability) than database. Can’t this be applied to traditional relational databases or any database ?
Would appreciate if you write another article detailing more on No SQL databases
Thanks again !
Thanks Ashish for an excellent quick summary of databases. Summarizing your article leaves below points:
1) Star Schema is used in OLAP systems, however an improvement can be made using column-column approach rather than traditional row-row approach. HP’s Vertica, SAP’s HANA and Amazon’s Redshift are examples.
2) Rather than having a database that does hard disk storage-memory writes and reads , thus wasting lot of time in that , one can opt for main memory and single threaded databases like Volt DB.
3) No SQL databases provides all together a different dimension of scalability, felxibility and supports various kinds of data models such as Cassandra, MongoDB and CouchDB.
Questions:
1. Can columnar databases be used in OLTP kind of systems ? If not ( not better suited) why ? If yes how ?
2. Main memory loading and single threading looks more like an charesteristic of a system ( or system architecture and RAM availability) than database. Can’t this be applied to traditional relational databases or any database ?
Would appreciate if you write another article detailing more on No SQL databases
Thanks again !
Bless the fitsttechnology is entering the market – I’d say Hadoop is the most well known example of the new entrants though not the first.
But I see far too many discarding the relevance of the RDBMS in favor of one or more of the new technologies
my name is Abdullahi