[Search for users] [Overall Top Noters] [List of all Conferences] [Download this site]

Conference ulysse::rdb_vms_competition

Title:	DEC Rdb against the World

Moderator:	HERON::GODFRIND

Created:	Fri Jun 12 1987
Last Modified:	Thu Feb 23 1995
Last Successful Update:	Fri Jun 06 1997
Number of topics:	1348
Total number of notes:	5438

880.0. "RMS files to RDB files" by AKOFIN::WOODRUFF () Thu Mar 07 1991 16:54

    	I am currently working for an internal group trying to put together
    a hardware capacity plan.  They currently run 15 production systems and
    all but 1 uses RMS files.  They will over the next 18 months be
    converting to new systems which use RDB files.  Since we don't have an
    unlimited supply of either memory or disk space I am trying to project
    how much space will be required to run these systems running RDB. 
    Does anyone have any information on this subject.  By the way the 1
    system that does use RDB files went from 50k blocks (RMS) to 200k
    blocks. (RDB)

T.R	Title	User	Personal Name	Date	Lines
880.1	No easy answer	TRCO01::MCMULLEN	Ken McMullen	`Thu Mar 07 1991 17:52`	19
	This is a question that is often asked and often not answered directly. The "shoot from the hip" answer is be prepared to provide 50-100% more disk. Why is that you ask .... you now have journal files, possibly snapshot files and a lot more functionality than you did with RMS which necessitates Rdb storing more information about the data. There are many notes in NOVA::RDB_40 conference on sizing databases. There are some tools posted that may assist you in your sizing efforts (see not 8 in NOVA::RDB_40). You should also read (several times) the Guide to Database Maintenace and Performance in the Rdb doc set. There are many things that will affect your space allocation. You need to know your application very well (ie read intensive, write intensive, future growth...). You may also consider using Rdbexpert to help you design your physical database. Good Luck, Ken M
880.2	Then there's normalization	BROKE::HIGGS	SQL is a camel in disguise	`Thu Mar 07 1991 23:41`	6
	Of course, if you do some level of normalization on the data you're converting (and I hope you do, or else you may find performance problems will rear their ugly head -- it isn't wise to just convert the RMS applications without looking at the design to see whether it is appropriate in a relational database approach), then it is possible that you will be able to eliminate some duplicate data, and reduce the extra overhead needed.