Comments on: Notes on the evolution of OLTP database management systems http://www.dbms2.com/2010/04/05/oltp-database-management-systems-2/ Choices in data management and analysis Thu, 09 Feb 2012 13:48:12 +0000 hourly 1 http://wordpress.org/?v=3.0.3 By: Joe Adinolfi http://www.dbms2.com/2010/04/05/oltp-database-management-systems-2/#comment-165053 Joe Adinolfi Fri, 09 Apr 2010 10:57:28 +0000 http://www.dbms2.com/?p=1841#comment-165053 Curt, i defintely agree that solid state memory is the future direction for analytical DBMS's. I have seen the slide ware on the Teradata Extreme Performance Appliance and if reality is even close then this technology is a game changer. There are questions on longevity and availability but since SSD's have no moving parts i would expect them to be more reliable than HD's. Pretty exciting stuff. Curt, i defintely agree that solid state memory is the future direction for analytical DBMS’s. I have seen the slide ware on the Teradata Extreme Performance Appliance and if reality is even close then this technology is a game changer. There are questions on longevity and availability but since SSD’s have no moving parts i would expect them to be more reliable than HD’s. Pretty exciting stuff.

]]> By: J. Andrew Rogers http://www.dbms2.com/2010/04/05/oltp-database-management-systems-2/#comment-164874 J. Andrew Rogers Wed, 07 Apr 2010 16:01:27 +0000 http://www.dbms2.com/?p=1841#comment-164874 Curt, I would be happy to say more offline, just drop me an email. Curt, I would be happy to say more offline, just drop me an email.

]]> By: Andy Ellicott http://www.dbms2.com/2010/04/05/oltp-database-management-systems-2/#comment-164742 Andy Ellicott Tue, 06 Apr 2010 12:51:41 +0000 http://www.dbms2.com/?p=1841#comment-164742 Hi Mark, Thanks for the kudos. BTW--we've opened up the VoltDB beta program and are providing more info about it. For those who want to try it or read the product documentation, just request access to the early release web site at http://www.voltdb.com. Hi Mark,
Thanks for the kudos. BTW–we’ve opened up the VoltDB beta program and are providing more info about it.

For those who want to try it or read the product documentation, just request access to the early release web site at http://www.voltdb.com.

]]> By: Curt Monash http://www.dbms2.com/2010/04/05/oltp-database-management-systems-2/#comment-164725 Curt Monash Tue, 06 Apr 2010 07:45:26 +0000 http://www.dbms2.com/?p=1841#comment-164725 Mark, You are very right to focus on TCO, as opposed to specific types of cost. CAM Mark,

You are very right to focus on TCO, as opposed to specific types of cost.

CAM

]]> By: Curt Monash http://www.dbms2.com/2010/04/05/oltp-database-management-systems-2/#comment-164724 Curt Monash Tue, 06 Apr 2010 07:44:12 +0000 http://www.dbms2.com/?p=1841#comment-164724 Andrew, Can you say more about that, online or off? CAM Andrew,

Can you say more about that, online or off?

CAM

]]> By: J. Andrew Rogers http://www.dbms2.com/2010/04/05/oltp-database-management-systems-2/#comment-164666 J. Andrew Rogers Mon, 05 Apr 2010 18:51:33 +0000 http://www.dbms2.com/?p=1841#comment-164666 Nova's comment likely refers to a new theoretical architecture for relational algebra implementations that is uniquely suited for massively parallelizing graph analytic databases. It will be difficult for an open source implementation to emerge because the core algorithms are buried in IP. There are at least two companies developing products based on it, one of which is a major database vendor. The middle ground for cases like this is to go straight to a cloud offering for cost-sensitive customers. Not ideal but workable and it allows the vendor to stay in control of their product. It is very difficult to make money open sourcing a product when core development is being funded entirely by the company. Nova’s comment likely refers to a new theoretical architecture for relational algebra implementations that is uniquely suited for massively parallelizing graph analytic databases. It will be difficult for an open source implementation to emerge because the core algorithms are buried in IP. There are at least two companies developing products based on it, one of which is a major database vendor.

The middle ground for cases like this is to go straight to a cloud offering for cost-sensitive customers. Not ideal but workable and it allows the vendor to stay in control of their product. It is very difficult to make money open sourcing a product when core development is being funded entirely by the company.

]]> By: Mark Callaghan http://www.dbms2.com/2010/04/05/oltp-database-management-systems-2/#comment-164649 Mark Callaghan Mon, 05 Apr 2010 14:16:48 +0000 http://www.dbms2.com/?p=1841#comment-164649 Many DBMS products are free to use but they aren't free to operate. Lots of people make a living keeping them running and the open-source products frequently require a bit more care than the closed- source ones. It would be interesting to read estimates on the point at which the open-source products have better price/performance. How big must a deployment be before the license cost from closed-source is much larger than the extra people cost from open-source? VoltDB and TokuDB (http://www.tokutek.com) are two of my favorites in the OLTP space. I wish they would write more about their products. Thanks for the link to the new H-Store paper. Many DBMS products are free to use but they aren’t free to operate. Lots of people make a living keeping them running and the open-source products frequently require a bit more care than the closed- source ones. It would be interesting to read estimates on the point at which the open-source products have better price/performance. How big must a deployment be before the license cost from closed-source is much larger than the extra people cost from open-source?

VoltDB and TokuDB (http://www.tokutek.com) are two of my favorites in the OLTP space. I wish they would write more about their products. Thanks for the link to the new H-Store paper.

]]>