Storage technologies for IBM Db2 Warehouse

A key decision that you must make for your Db2® Warehouse deployment is the type of storage technology to set up and use.

Db2 Warehouse can run multiple parallel engines. Each engine accesses independent file sets for an optimized, shared-nothing pattern that helps provide high performance.

A Db2 Warehouse MPP deployment requires a POSIX-compliant cluster file system, which provides servers and other resources with concurrent access to a single file system. Examples of a POSIX-compliant cluster file system are IBM Spectrum Scale (formerly IBM® General Parallel File System, or GPFS) and VxFS, which you mount on /mnt/clusterfs, as shown in Cluster file system.
Figure 1. Cluster file system
Choose the storage technology that best satisfies your business requirements and constraints, for example:
  • Cost. How much does it cost to purchase and operate?
  • Scaling capabilities. How fast and easy is it to add or remove resources as needed?
  • Ease of operation. Is it easy to set up and administer?
  • Performance. How fast can it access data?
  • Protection. Does it keep your data secure?
  • Attachment method. Should the nodes be attached a network-attached storage (NAS) device or to a storage area network (SAN)? For more information, see Table 1.
Table 1. Comparison of NAS versus SAN
  Network-attached storage (NAS) Storage area network (SAN)
Description NAS is a dedicated file storage device that is connected to a network that allows data to be stored in and retrieved from a central location. A SAN is a dedicated network that connects all storage devices so that they work as a single storage block.
Ease of deployment NAS is easier to deploy on an existing architecture. Deployment of a SAN requires more knowledge of communication protocols and extra cluster file system configuration.
Performance NAS is usually slower because it typically shares data within the communications network. A SAN provides excellent speed and performance because it is a dedicated storage network.
Data access NAS provides file-level data access. A SAN provides block-level data access.
Protocols Files are shared and stored by using standard file system protocols, mainly NFS. Communication protocols for a SAN include Fibre Channel and iSCSI.
Deployment of a Db2 Warehouse container In a NAS environment, because the storage is already configured, deploying a Db2 Warehouse container is straightforward. In a SAN environment, you must configure storage devices to deploy a Db2 Warehouse container.
Other traits NAS offers ease of use and management. The ability to scale capacity is one of the main reasons for using a SAN.