RE: OT RE: Async I/O on Windows - WHAT is a FEDERATED DATABASE

From: Tony Johnson <tjohnson_at_griddata.com>
Date: Tue, 06 Feb 2001 09:51:40 -0800
Message-ID: <F001.002AC205.20010206081054@fatcity.com>

The
whole idea behind 9i is CacheFusion which uses a high-speed <SPAN
class=321141016-06022001>interconnect to solve the pinging issues. At least that is the marketing
line
that will only be proved in time. Any database of any size should
be
using partititioning if you want it to perform and be able to manage it.
<FONT

size=2>------------------------------------------------------------------------------Tony 

Johnson           

        
        
        Email  : 

tjohnson_at_griddata.comSenior Database Administrator          
        Voice  : ( 480 ) 682 -
0849GridData       

        
        
        
        Cell   : ( 602 ) 363 - 
7328             
       7408 W. Detroit 
#100            

       
        Fax    : ( 480 ) 961 -
8801Chandler, AZ

85226------------------------------------------------------------------------------Murphy's 

Data Constant:Data will be damaged in direct proportion to its value

<FONT face=Tahoma

  size=2>-----Original Message-----From: root_at_fatcity.com   [mailto:root_at_fatcity.com]On Behalf Of Mohan, RossSent:   Tuesday, February 06, 2001 6:53 AMTo: Multiple recipients of list   ORACLE-LSubject: RE: OT RE: Async I/O on Windows - WHAT is a   FEDERATED DATABASE
  I
  understand the argument, Rodd and it raises three   points/questions:
<FONT color=#0000ff face=Arial

  size=2> 

1. I can always back up a "state" ( part of a federation?) just like EMC/SRDF/BFD SAN does
   <FONT color=#0000ff face=Arial
   size=2>    for the Oracle solution, and at less cost, and
   <FONT color=#0000ff face=Arial
   size=2>  2) Do you believe you can simply "add nodes" to an OPS farm to improve performance? I have     <FONT color=#0000ff face=Arial size=2>personally never gone over a humble two nodes in OPS, and even then, locking issues must     <FONT color=#0000ff face=Arial size=2>be addressed. One way out of this is the geographically segregate and partition the data. But
   <SPAN
   class=353224813-06022001><FONT size=2><FONT color=#0000ff face=Arial>    this would be "federated."  In a pure play OPS scenario, I would imagine the system would
   <SPAN
   class=353224813-06022001>    <FONT color=#0000ff face=Arial>choke to death after the fourth or sixth node, without special tweaks like partitioning, either
   <SPAN
   class=353224813-06022001>    by data or application.
   <SPAN
   class=353224813-06022001> 
   <SPAN
   class=353224813-06022001>3) Loss of a SS "state", just like loss of an oracle partition, does not "kill the operation of the system".

<SPAN

  class=353224813-06022001>    here, they are the same.   ......
<SPAN

  class=353224813-06022001> 
<SPAN

  class=353224813-06022001>just a thought......
<SPAN

  class=353224813-06022001> 
<SPAN

  class=353224813-06022001> 
<SPAN

  class=353224813-06022001>   
<SPAN

  class=353224813-06022001> 
<SPAN

  class=353224813-06022001> -----Original   Message-----From: Holman, Rodney
  [mailto:rodney.holman_at_lodgenet.com]Sent: Tuesday, February 06, 2001   5:21 AMTo: Multiple recipients of list ORACLE-LSubject:   RE: OT RE: Async I/O on Windows - WHAT is a FEDERATED   DATABASE        <SPAN
    class=470590510-06022001>Ross,
    I
    was at the Open World conference session where Jeremy Burton made the     comments about clustering, OPS, data segmentation, etc.  The data     segmentation part was about MS SQLServer, and about how it creates     significant work to add cluster nodes. C|net has their terms and     comments a little scrambled. The Oracle 9i solution used OPS for the     instances and an EMC/SRDF SAN for the data storage.  Each OPS cluster     node had full access to every piece of data.  By doing this no node is     a single point of failure (as Larry demonstrated and was chastised for by     MS).  Also it creates greater capability for scalability.  Just     configure and add a node and it improves performance (also part of Larry's     demo).  As described with the MS federated database configuration you     would need to resegment the data to add a node.  This would then     destabilize the system even further by adding another single point of     failure.  Failure of an OPS cluster node with the data on a SAN      with redundancy, such as the EMC/SRDF option, only decreases performance, it     doesn't kill the operation of the system.     <SPAN
    class=470590510-06022001> 
    <SPAN
    class=470590510-06022001>Rodd Holman     

      <FONT face=Tahoma 
      size=2>-----Original Message-----From: Mohan, Ross 
      [mailto:MohanR_at_STARS-SMI.com]Sent: Monday, February 05, 2001 
      5:09 PMTo: Multiple recipients of list 
      ORACLE-LSubject: RE: OT RE: Async I/O on Windows - WHAT is a 
      FEDERATED DATABASE
      Very Interesting!  It appears Oracle 9i, is, in fact, 
      a Hybrid Federated Database! 
      <A 
      href="http://news.cnet.com/news/0-1003-200-2897140.html?tag=st.ne.ni.metacomm.ni" 
      target=_blank>http://news.cnet.com/news/0-1003-200-2897140.html?tag=st.ne.ni.metacomm.ni 
      
      A snippet: 

Received on Tue Feb 06 2001 - 11:51:40 CST