Anastassia Ailamaki: Staged Database Systems

GOOG being what it is and computing moving away from the 70ies, 80ies, 90ies "Client/Server" computing and "HFS/RDB" (Hierarchical File system/Relational Database) storage paradigm, I still think there are lots of applications for small computers and more conventional research. Quite a lot of people still do good work in those fields.

One of them seems to be: Anastassia Ailamaki.

And two of the fields she works in are:

Staged Database Systems: The main philosophy of the Staged DB design is to group concurrent tasks per software and hardware resource in the system. We do not propose a major overhaul in existing DBMS software. Rather, a small number of targeted changes can encapsulate existing functions into micro-servers (stages) ready to process a group of tasks. Papers are here.

Data Mining meets Traffic Modeling Traffic modeling of storage workloads is extremely helpful in evaluating system designs. They got a nice abstract (pdf) too there.

So, if these people are really good, relieve might come to us poor asses with lots of content in RDBMses.

My own ideas of "staged databasing" were a lot more primitive: basically I had the ideas to run fresh content from small mysql dbs and servers and automate backending in large Oracle stores or so. Probably hard to do & not very efficient. Sometimes one should listen to science a bit more and study a problem in earnest.

Apropos database features and cost: A Comparison of Oracle, MySQL and PostgreSQL DBMS from the Fermi National Accelerator Lab.

Jeremy Zawodny asks How Cheap is MySQL? SPEC gives a classical answer. mySQl set a TOPSC (TotalOperationsPerSecond Cost = System Cost divided by TOPS) record of $82 in 2004 but in the meantime the large ROI companies have came back to fight and the record stands at $47 as of first quarter 2005 The new record holder is SQLServer on Windows Server 2003 running HP (Proliant HW) running . The cheapest Oracle solution comes on Dell Itanium machines running 10g on Linux at $101.

On the other hand the Oracle solution came out at 1656 TOPS, mySQl on Sun renders 1363 TOPS and the priceworthy HP/MS DBMS at 1001. Of course these tests are highly abstract and do not consider TCO (Total Cost of Ownership) or anything daily real world. What we can see though, is, that overall competition is tight. Probably the most important factor is: where did you invest your team's learning energy?

Accidental discovery of today: What used to be Red China is called Mainland China now.

... plink