Storage Foundation and High Availability Solutions 7.4.2 Solutions Guide - Windows
- Section I. Introduction
- Introducing Storage Foundation and High Availability Solutions
- Using the Solutions Configuration Center
- SFW best practices for storage
- Section II. Quick Recovery
- Section III. High Availability
- High availability: Overview
- How VCS monitors storage components
- Deploying InfoScale Enterprise for high availability: New installation
- Notes and recommendations for cluster and application configuration
- Configuring disk groups and volumes
- Configuring the cluster using the Cluster Configuration Wizard
- About modifying the cluster configuration
- About installing and configuring the application or server role
- Configuring the service group
- About configuring file shares
- About configuring IIS sites
- About configuring applications using the Application Configuration Wizard
- About configuring the Oracle service group using the wizard
- Modifying the application service groups
- Adding DMP to a clustering configuration
- High availability: Overview
- Section IV. Campus Clustering
- Introduction to campus clustering
- Deploying InfoScale Enterprise for campus cluster
- Notes and recommendations for cluster and application configuration
- Reviewing the configuration
- Configuring the cluster using the Cluster Configuration Wizard
- Creating disk groups and volumes
- Installing the application on cluster nodes
- Section V. Replicated Data Clusters
- Introduction to Replicated Data Clusters
- Deploying Replicated Data Clusters: New application installation
- Notes and recommendations for cluster and application configuration
- Configuring the cluster using the Cluster Configuration Wizard
- Configuring disk groups and volumes
- Installing and configuring the application or server role
- Configuring the service group
- About configuring file shares
- About configuring IIS sites
- About configuring applications using the Application Configuration Wizard
- Configuring a RVG service group for replication
- Configuring the resources in the RVG service group for RDC replication
- Configuring the VMDg or VMNSDg resources for the disk groups
- Configuring the RVG Primary resources
- Adding the nodes from the secondary zone to the RDC
- Verifying the RDC configuration
- Section VI. Disaster Recovery
- Disaster recovery: Overview
- Deploying disaster recovery: New application installation
- Notes and recommendations for cluster and application configuration
- Reviewing the configuration
- About managing disk groups and volumes
- Setting up the secondary site: Configuring SFW HA and setting up a cluster
- Setting up your replication environment
- About configuring disaster recovery with the DR wizard
- Installing and configuring the application or server role (secondary site)
- Configuring replication and global clustering
- Configuring the global cluster option for wide-area failover
- Possible task after creating the DR environment: Adding a new failover node to a Volume Replicator environment
- Maintaining: Normal operations and recovery procedures (Volume Replicator environment)
- Testing fault readiness by running a fire drill
- About the Fire Drill Wizard
- Prerequisites for a fire drill
- Preparing the fire drill configuration
- Deleting the fire drill configuration
- Section VII. Microsoft Clustering Solutions
- Microsoft clustering solutions overview
- Deploying SFW with Microsoft failover clustering
- Tasks for installing InfoScale Foundation or InfoScale Storage for Microsoft failover clustering
- Creating SFW disk groups and volumes
- Implementing a dynamic quorum resource
- Deploying SFW with Microsoft failover clustering in a campus cluster
- Reviewing the configuration
- Establishing a Microsoft failover cluster
- Tasks for installing InfoScale Foundation or InfoScale Storage for Microsoft failover clustering
- Creating disk groups and volumes
- Implementing a dynamic quorum resource
- Installing the application on the cluster nodes
- Deploying SFW and VVR with Microsoft failover clustering
- Part 1: Setting up the cluster on the primary site
- Reviewing the prerequisites and the configuration
- Part 2: Setting up the cluster on the secondary site
- Part 3: Adding the Volume Replicator components for replication
- Part 4: Maintaining normal operations and recovery procedures
- Section VIII. Server Consolidation
- Server consolidation overview
- Server consolidation configurations
- Typical server consolidation configuration
- Server consolidation configuration 1 - many to one
- Server consolidation configuration 2 - many to two: Adding clustering and DMP
- About this configuration
- SFW features that support server consolidation
Striping for data-transfer-intensive applications
The ideal stripe unit size for data-transfer-intensive applications that use a striped volume is the typical I/O request size of the application, divided by the number of data disks in the stripe. For example, if an application typically makes requests for 256 KB, an ideal stripe size for a four-disk striped volume would be 64 KB (256 KB/4).
Data-transfer-intensive applications typically request a large amount of data with every request, between 64 KB and 1 MB, or more. When a large amount of data is requested, the data-transfer phase of the request represents the majority of the request execution time. Thus, reducing data-transfer time improves I/O performance.
A single disk can transfer data only as fast as the data passes under the disk's read-write head. For example, a disk that rotates at 10,000 RPM and has 200 blocks on a certain track cannot transfer data to or from that track any faster than 17.06 MB per second (200 blocks x 512 bytes per block/0.006 seconds per revolution). An application request for 500 KB would require five platter revolutions, or 30 milliseconds, to execute. If the request were addressed to a volume of five identical disks created with SFW, each disk would ideally deliver one-fifth of the data, and the request would complete in a shorter time.
In general, if a striped volume is optimized for data-transfer-intensive applications, each application I/O request will split evenly across all of the volume's disks (or all but the disk containing parity data in the case of a RAID-5 volume).