Scalability architecture: SMP and MPP


SMP stands for symmetric multi-processing, distribution of computingworkload across a cluster of linked computers. Each node may havemultiple, multi-core processors to further parallelize processing on thatnode. Nodes may have local disk storage and also share disk storage withother nodes. A database system that runs across nodes of an SMP clusteris termed “shared disk.” It uses locks to regulate concurrent access to thedatabase, although locking is typically not a bottle-neck for read-intensivedata-analysis applications.MPP stands for massively parallel processing. MPP enables a “sharednothing” database architecture where each computer-node has only localmemory and disk storage. The database is partitioned across the nodes ofthe MPP cluster, which is termed a “grid” when the hardware is homogeneous.

Sybase IQ uses SMP for database scale-up/out through the addition ofprocessors or compute nodes with the additional option of multiplex gridtechnology to support node scale-out. It is possible to assign differenttasks to compute nodes that are optimized for those tasks. Computeresources and storage scale independently.

Sybase IQ Analytics Server- Competitive Assessment




  • auctio

    I do not even fully grasp how I stopped up correct here, however I thought this put up was wonderful. I do not recognize who you may be but absolutely you are going to a renowned blogger within the event you aren’t already. Cheers!

Leave a Comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.