1 / 15

How to Cluster both Servers and Storage

How to Cluster both Servers and Storage. W. Curtis Preston President The Storage Group. What is a Cluster?. Two or more servers that behave like one Purists will say it should be several servers Many “clustering” products have supported only two members of a single cluster Active/passive

junius
Télécharger la présentation

How to Cluster both Servers and Storage

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. How to Cluster both Servers and Storage W. Curtis Preston President The Storage Group

  2. What is a Cluster? • Two or more servers that behave like one • Purists will say it should be several servers • Many “clustering” products have supported only two members of a single cluster • Active/passive • Standby resource is not used until failure occurs • Active/active • All resources in cluster share the application load • Failure of one node increases the load on remaining nodes • Node resumes its role if it returns to cluster

  3. Why Cluster? Application Availability • Goal is to ensure that application is always available, even if a cluster member is down • Active/passive: application continues to function if standby server is available • Active/active: application continues to function if at least one member of cluster is available

  4. Why Cluster? Performance and Scalability • The load is distributed across multiple servers • Applications can perform faster than on any single server • Large I/O requests (For example, Web farms) • Large database queries and processing (For example, parallel database) • Applications can scale farther than on any single server • Can increase capacity by adding node to the cluster • Capacity can be added without stopping application

  5. Examples of Clustering • VAX • IBM S/390 • SP/2 • VERITAS Cluster Server • MS Cluster

  6. Shared Storage Clustering • Storing function • Shared SPI (parallel SCSI) • Possibly other bus topology • Filing function • Non-shared file systems • Serving function • Clustering was done here • Clustering application responsible for locking data and time-sharing write access to disk

  7. Independent Storage Clustering • Storing function • Non-shared storage, even in fail over • Filing function • Non-shared file systems • Files and file systems mirrored between nodes using software • Serving function • Clustering still done here • No longer need to time-share write access to disk • Increases availability • Either node can service application via its own mirrored copy of the data

  8. Storage Can be Clustered Too! • To cluster our storage, we need multiple resources acting as one virtual resource • Can push the purposes of clustering to the storage level • Increase the capacity of the storage system by distributing the load • Increase the availability of the application by increasing entire system’s ability to • Sense • Report • Recover

  9. Clustering the Storing Function • Two or more disks, virtual disks, or arrays can be made to appear as one • Functionality provided by • SAN-aware Volume Managers • I/O based mirroring • Intelligent routers • Can fulfill performance, scalability, and availability purposes of clustering • Note: Filing function not clustered

  10. Clustering the Filing Function • Clustering on the backend • Allows multiple machines to access the filesystem as if they were one machine. • Software required on each server that will access filesystem • Several proprietary products available today • DAFS is a future, non-proprietary filesystem that should offer this functionality • Note: Storing function not clustered

  11. Best Practice Clustering Storing and Filing with SAN • Combine a single, clustered, virtual volume with a single, clustered virtual filesystem • Clustering end to end • Load distributed between all available servers, disks, and SAN components • Any server, SAN component, or disk array, is backed up by at least one other

  12. Best Practice Clustering Storing and Filing with NAS • NAS vendors also offer a complete clustered solution • NFS or CIFS already a multiple access filesystem • Also uses clustered storage on backend • Can offer similar levels of availability, performance, and scalability as a SAN-based product, with a proven file system

  13. Future Developments • Distributed filesystems • Existing products should become more mature • New products will drive competition • DAFS • Will bridge the gap between SAN and NAS • iSCSI • Allow for use of traditional network management tools in SAN • Could also use OSPF instead of FSPF

  14. Clustering is a Good Thing • Clustering a system’s serving, filing and storing functions results in a completely clustered, load-balanced, high-available system • There is more than one way to do it! • Fibre Channel SAN • iSCSI SAN • NAS with NFS/CIFS • NAS with DAFS

  15. Learn more… • Directory of vendors offering storage clustering options:http://www.storagemountain.com • Using SANs and NAS now available from O’Reilly & Associates • Questions to:curtis@thestoragegroup.com

More Related