IBM Spectrum Protect LAN-free integration with Netezza Performance Server

IBM Spectrum® Protect (formerly known as Tivoli® Storage Manager) is a 3rd party platform, which you can use to manage Netezza® Performance Server database backups. This capability is available with Cloud Pak for Data System 1.0.8.3 and later.

Netezza Performance Server acts as a client of the IBM Spectrum Protect server. Database backups or restores can be executed with nzbackup or nzrestore command with suitable value set for -connector argument (for IBM Spectrum Protect, the value is tsm). The backup payload is stored on media devices that are managed by the IBM Spectrum Protect server. In the most common scenario, the data flows over the Ethernet network.

IBM Spectrum Protect provides special capabilities to take backups directly on SAN. In this scenario, data flows over the Fibre Channel storage area network and do not impact throughput of the Ethernet network. In addition to the Ethernet network offloading, this approach might also improve the overall backup execution time. This capability is called LAN-free backup and it supports two types of SAN storage:
  • Tape library (or virtual tape library)
  • Spectrum Scale - shared file storage (GPFS)

In addition to the specific SAN configuration, LAN-free infrastructure is based on a storage agent (SA). Storage agent is an IBM Spectrum Protect component, which runs on client side and is used to store the data.

For IBM Spectrum Protect LAN-free integration prerequisites, see IBM Spectrum Protect LAN-free integration prerequisites.

See the IBM Spectrum Protect documentation for general information about the LAN-free configuration:

Target audience for IBM Spectrum Protect LAN-free integration

Consider IBM Spectrum Protect LAN-free capabilities if:
  1. You are using Netezza Performance Server.
  2. You want to manage database backups within IBM Spectrum Protect server.
  3. You want to improve backup execution time or separate the backup data traffic between Ethernet and SAN FC network.
If you are a high-level user and want to implement IBM Spectrum Protect LAN-free solution, you need to have the following elements:
  • Cloud Pak for Data System 1.0.8.3 or later with at least one connector node (although two Connector Nodes are recommended).
  • IBM Spectrum Protect server license and a Linux® machine with FC connectivity on which the server can be deployed.
  • Additional external IP addresses, which are deployed on connector nodes.
  • SAN infrastructure that is attached to CPDS Connector Nodes and IBM Spectrum Protect server node.

High-level integration architecture

To implement IBM Spectrum Protect LAN-free for Netezza Performance Server, which is deployed on CPDS 1.0.8.3 platform, you can used both supported SAN storage types (tape library or Spectrum Scale). Tape library (or VTL) was not tested with CPDS 1.0.8.3 so far, but there should be no obstacles to use it other than special storage device availability. The integration that is described in this section focuses on IBM Spectrum Scale (formerly GPFS) storage.

IBM Spectrum Protect server is installed on external Linux machine (or a cluster of machines). IBM Spectrum Scale cluster and common file system are created between CPDS connector nodes and external Linux machine. IBM Spectrum Scale file system is based on a set of SAN devices, which are exposed to all the nodes in the cluster. It means that CPDS connector nodes and external Linux machine see the same set of external SAN disks. The file system is mounted to all the cluster nodes. In this way, IBM Spectrum Protect server can manage the volume files that are stored on common file system. At the same time, Netezza Performance Server can store data directly (LAN-free) to the volume files. This is where IBM Spectrum Protect client and storage agent components are used.

The platform already uses IBM Spectrum Scale cluster, which maintains ips and platform file systems across the control plane and connector nodes (nodes on which Netezza Performance Server host container can run). In IBM Spectrum Scale LAN-free integration, separate IBM Spectrum Scale is created. It means that once the integration is in place, there are two clusters within CPDS platform:
  • Internal CPDS cluster that is used to manage ips and platform file system; defined on CPDS nodes and disks only (control plane nodes).
  • External cluster that is used to manage backup (tsm) file system; defined on CPDS nodes (connector nodes) and non-CPDS nodes (IBM Spectrum Protect server node) and external SAN disks.

To define external IBM Spectrum Scale cluster between CPDS connector nodes and IBM Spectrum Protect server node, it is needed to start the additional external IP addresses on CPDS connector nodes. These addresses are used by IBM Spectrum Scale software for the direct communication between the nodes. It is recommended that additional external IP addresses are from the same network as IBM Spectrum Scale server IP and Netezza Performance Server application IP.

To learn more about your responsibilities and IBM responsibilities, see Responsibility matrix.