Oracle FAQ | Your Portal to the Oracle Knowledge Grid |
Home -> Community -> Usenet -> c.d.o.server -> Re: ORACLE vs INFORMIX vs MS SQL SERVER
Didier has it right (in my opinion). For that amount of data you really
have to (consider) seperating the image data from the rest. This is pretty
much a standard way of storing such a large volume of images, it is used a
lot by archiving and imaging type organisations. Probably the biggest issue
will be what to store that many large images on. I have seen a couple of
things done, such as large optical drive carousels, but they are quite often
slow and only allow a single user to access a file at a time. Something I
have only played with but you might want to have a look at is a Netapp
filer.
You can visit the appropriate web site at http://www.netapp.com. Netapp
filers are just about the fastest thing around, scalable, will operate with
either NT or UNIX and are the only file server officially endorsed by
Oracle. They are being used in a lot of particularly storage intensive
industries like music and video production and to serve large web sites etc.
Regards
Rod J. Stewart
Didier Quesson wrote in message <726lh3$tod$1_at_minus.oleane.net>...
>For a so large amount of data, SQL storage may not be appropriate, while
>native support by filesystems would be more easy to figure out. In that
case
>you need a very good file server solution, and to integrate a database that
>provides links to the actual images on the appropriate filesystems.
>That way, the RDBMS database will be kept small and more easily optimizable
>for fast indexing of this document database, used for searches and so on.
>You should really consider separating image storage (may be on multiple
file
>servers) from the RDBMS itself, by using hyperlinks to access them from the
>database. Then your SQL queries will not retrieve the image data itself but
>the hyperlinks to the image data files.
>I suggest a dedicated server for the Web server and the RDBMS, plus
multiple
>file servers to store the actual data. Management of this large database of
>files may be integrated within the RDBMS data model (which will hold the
>whole directory of files, but not the data).
>Data retrieval by the Web server may be operated by CGI or ActiveX scripts
>which will make search in the database using SQL queries to perform
searches
>in the directory tree, and then will collect the hyperlinks stored in the
>database to process image data retrieval down to the client. Such a
solution
>is then very extensible, because it does not limit the number of file
>servers which will process file data retrievals.
>
>
Received on Mon Nov 09 1998 - 16:52:27 CST