MySQL Cluster - Highly Reliable, Speedy Database Solution

Replies

• 

  PraveenKumar Purushothaman

  Seems promising... I guess Oracle DB is gonna be replaced by MySQL, but why do they do that?
• 

  Kaustubh Katdare

  The Mysql Cluster isn't a dig at the Oracle; as far as I'm aware. They're talking about a new technology, I guess. Looking forward to responses from CS engineers.
• 

  PraveenKumar Purushothaman

  Let me get some more information from my friend at Oracle... 😀
• 

  Kaustubh Katdare

  Can we expect more responses from compute science engineers here? 😀
• 

  Prashant Munshi

  At outset let me confess that I'm not a CS engineer by academic degree but have been working in the field of databases for some 17+ years, so taking this privilege to answer this question.
  
  The lexical meaning of the word cluster is "grouping of related items held together" and that's what it also applicable to databases. There are two types of database clustering commercially available - Shared Nothing (SN) and Shared Everything (SE). The MySQL clustering falls in the former (SN) of these two categories, whereas Oracle's RAC falls in the second category (SE), though former is also possible with Oracle, in which case they call is as "distributed databases". Actually Shared Nothing by far is distributed database and it has several databases each running on different machine (called nodes), independently but networked together, with each database is complete in its own right, containing the tables of data relevant to the local context, but for privileged users use these discrete databases logically in the form a single large database, transparent to the user that he / she is actually getting the data from different databases. You may visit #-Link-Snipped-# for a more elaborate description of DDBMS.
  
  However the Shared Everything architecture consists of multiple (not databases) instances running on multiple different servers mount a common database from "shared disk subsystem". Shared Disk Clustered Database Architecture run on hardware clusters that give every participating server equal access to all disks. However, memory is not shared. Each machine runs an "instance" of database which is technically defined as combination of shared memory and database processes, which is prime mechanism to support the data maneuvering by the users. Oracle's Real Application (RAC) technology, introduced in Oracle 9i in 2004 was the first commercially successful implementation of clustering with "cache fusion". Since the explanation is long, please read further description from PDF and post your doubts here which may then be easy for me to offer point wise explanation.