This content has been marked as final. Show 7 replies
Do you have requirements in place ?
Does you system is ready (development system) ? If no, then certainly, building production system as such with requirements would not carried out. Since, atleast some tiny front end application would get tied up. So, try to get the requirements and check with developers (from their perspective). So that you can get and view point of requirement scope. Try to build an databas for development, as days goes on your requirement (issues day to day) would make the things classify the storage you need opt for. So that is how building any productions goes across AFAIK.
You can wait for inputs from other members of forums too.
- Pavan Kumar N
yes of course There's no one size fit all solution , but really we're looking for the best similar cases , we are at analyzing phase so we can consider all the possibility to find the best solution , really all our subscribers will querying this DB for checking their status and their consumed volume via another DB link query , we're little bit confused of using Data Guard or RAC or Linux cluster , of course we'll using interval partitioning
And what is the best designing for our DB to be ready for this huge Data with the best performance and with reasonable query time .
876602 wrote:I assume that by design, you imply the database architecture to use?
please I need your suggestion of the best design of how millions of data are gathered on a daily basis, stored in tens of on-line tables and then delivered to the user’s desk in real time for strategic decision making ,
I suggest looking at Oracle RAC. Oracle's RDBMS cluster solution.
H/w is the critical component that makes RAC scale. So it is very important to get that right.
Oracle Database Appliance is a low cost 2 node RAC h/w solution - with h/w, Interconnect, storage and o/s certified for Oracle RAC. Database Appliance is also not only cheap h/w wise (very competitively priced), there are zero overheads ito time and resources to prep and configure the h/w for RAC.
The downside is that Database Appliance does not scale up. You cannot expand this cluster by adding more nodes.
If that is an issue, the next option is Oracle's Exadata Database Machine (quarter, half or full rack). This provides high-end performance and scalability. Again, there are zero overheads in prepping and configuring the h/w for used as a cluster.
The alternative is to build the cluster yourself. This can be a complex exercise as you are then responsible for ensuring that the different parts work together, are correctly setup and configured, and so on. If you do not have any cluster experience, there is a steep learning curve and an increased risk of things going wrong.
FWIW, our largest cluster (self build) is pretty much used for similar business solution as you described. It handles over 4 billion row inserts per day, and processes these into various data models and reporting layers, for business use. H/w wise and Interconnect wise, it is a bit dated - but even so, manages to deal with daily data processing volumes with capacity to spare.
Oracle RAC is an excellent architecture and pretty much unique within the RDBMS world. I have yet to see anything else clustering wise, that can provide any real competition ito scalability, availability and redundancy.
I think you must read [url http://docs.oracle.com/cd/E11882_01/server.112/e25555/toc.htm] Data Warehousing Guide . Just give time to read it thoroughly to get into basics and then match with your environment, needs, available resources and business data nature. Apart from docs below book may also be of your interest :
Oracle DBA Guide to Data Warehousing and Star Schemas ISBN: 0130325848
And if this forum don't helps you more than there is separate forum too.
Below two links may give you more ideas :
Re: How to design a DataWareHouse
Re: Sugesstions to build Datawarehouse Environment