If I was planning on choosing a database management system and know that the data needs to be available as close to 100% as possible and that it needs to scale as well as possible, is there a "best" database management system for that at this time? This assumes that I am not looking at other factors such as cost. I'm having such a hard time comparing different DBMS (such as Oracle, Microsoft SQL Server, MySQL) and looking to see if there is a single one that outperforms the other. I think an answer is because there is no research that exists to do this comparison. Does anyone know otherwise? Thanks.
Choosing a database management system to maximize availability and scalability [closed]
1You can't say if one DBMS is faster than the other just like that. It totally depends on the workload, the type of queries and requirements you have. Do you need transactions? Do you need multi-master mode or even a clustered setup? Do you need hot backups? etc.– gertvdijkDec 14, 2012 at 19:40
Shopping Questions are Off-Topic on any of the Stack Exchange sites. See Q&A is hard, lets go Shopping and the FAQ for more details.– Chris SDec 14, 2012 at 19:40
If there were a "best" one, there wouldn't need to be any other ones. You haven't listed all of your requirements, particularly what your budget for this is. Any of the ones you named can be configured with HA clustering to ensure uptime. "Scale as well as possible" is pretty vague; it really depends on what problem(s) you're solving by "scaling".– mfinniDec 14, 2012 at 19:41
is there a "best" database management system for that at this time?
Oracle Exadata. Hands down. We used the last generation in my second last project on a 3 instance full autoamtic replaction system with each insatnce having5 mondes, 21.000gb of storage and it flew- that was the smallest setup they sold, btw., cost more than a quarter million USD per instance. We used that with a 100% uptime requirement (5x9 - 99.999) and double redundancy did it, especially as the replication was over 10km in 2 data centers with a 10g fiber link dedicated to the database server.
You can easily scale that to 100.000gb with storage nodes reaching a million IOPS per storage node with an all SSD configuration. THe price will approach a million per instance then. ORacle sold like 1000 systems last year, we took 3.
The reason they fly is that they have some ery non-standard hardware (infiniband - still new in many data centers) and are selling optimized hardware to go with the software - super appliance, and "money does not matter".
I think an answer is because there is no research that exists to do this comparison.
Ah, no, there is a LOT of data no that. Ther are standardized performance tests for that.
So, it's the best if your budget is unlimited, you're saying?– mfinniDec 14, 2012 at 19:40
Actually no. Hat would be DB2. Even exadata does not even come close to the scalability of a big iron mainframe cluster. I Remember something about 28 nodes in a cluster, with every node being a separate mainframe (which is itself consisting of a LOT of nodes). This gets into double digit millions though. Check publib.boulder.ibm.com/infocenter/dzichelp/v2r2/…– TomTomDec 14, 2012 at 19:42
Are these standardized performance tests published online where anyone would be able to see how a certain DBMS performs? Thanks!– O_ODec 14, 2012 at 19:42
1Ys, but you have a problem - at this sizes the db is not relevant, those testss test complete systems. Because you have problems getting dfata in / out, makineenough requsts. Check TPC-C in google.– TomTomDec 14, 2012 at 19:44
So, DB2 on zSeries, not just DB2. That's a whole platform you're recommending, not just a DBMS.– mfinniDec 14, 2012 at 21:32