Site survey and evaluation
Your IBM® Netezza® representative will work with you to evaluate your existing or chosen Netezza systems and environment. The site survey questions in this topic can help plan for the implementation of a replicated environment.
The questions are as follows:
- Is this a new or established Netezza environment? If new, your IBM Netezza representative will work with you to plan the Netezza systems and the hardware that your environment will require. Proceed to question 2. Otherwise, if you already have Netezza systems, review the following questions:
- What are the types and current releases of the existing Netezza systems? For existing customers, a replication environment can include different appliance types and model sizes, but the systems must run the same Netezza software release. The Netezza systems must also run Red Hat Enterprise Linux with NFS features and Netezza Firmware, Diagnostics, and Tools (FDT). For model and NPS® operating system requirements, see Table 1. For replication log server hardware and software requirements, see Table 1.
- What are the available interface cards and open ports on the Netezza systems? The replication solution requires two 10 GbE ports on each Netezza high availability host for redundant connections to the replication log server. The ports should be on different network interface cards (NICs) to ensure communication if one NIC fails. If the host does not have enough free ports, additional interface cards might be required. If the host does not have open PCI slots for additional cards, your IBM Netezza representative will work with you to identify possible alternatives.
- Are there systems that can serve as master and subordinates, or will another Netezza system be required? If a second Netezza system is available to serve as the subordinate node, it must be one of the supported IBM Netezza or IBM PureData™ System for Analytics appliances (1000, N1001, N2001, or C1000). The system must also meet the software release and interface prerequisites for its server host connection. A master and subordinate can support other, non-replicated Netezza database activity in addition to the replication activity.
- Do the systems use software or a service to ensure system time synchronization? The NPS hosts and the replication log server hosts must use time synchronization software to ensure that the system times are synchronized, which ensures a consistent time state for commands and changes in the replication environment.
- Where will the additional hardware be located? Any physical server, network switch, or array hardware must be installed in a separate rack (not the Netezza system rack) within the data center. To identify the cabling requirements for your environment, work with your company’s IT infrastructure team or your IBM Netezza representative to determine the wire length from the NPS hosts to the replication log server host.
- What are the expected sizes of the replicated databases and the frequency of the common load and insert operations against the replicated databases? Try to approximate the sizes of the replicated databases and the frequency of the loads and inserts into those databases. This information can help you to assess the WAN connection bandwidth and the amount of storage that is needed in the replication storage array. For example, when you load or insert data into replicated databases, the data is temporarily stored on the replication log servers and then transferred to update the subordinates after the master is updated. When you activate a new subordinate node or recover a replicated database, the backup of the replicated databases on the master can optionally be staged on the log server and automatically transferred to the subordinate. This optional capability is suitable only when the log server has available disk space to sufficiently stage the entire backup for the length of time that is required to transmit and restore to the subordinates. Be aware, however, that the amount of raw data does not determine the final value. For more information, see WAN bandwidth requirements.
- What are the bandwidth and configuration of the WAN connection between the replication log servers? The replication feature requires a minimum of 100 Mb of bandwidth. However, you should attempt to calculate how much data you will be replicating. If that isn't possible, overestimate the needed bandwidth. In addition, it is highly recommended that you consult your network administrators to identify VPN or other solutions in your environment to secure the traffic over the WAN link. Security solutions vary, and some can incur a performance overhead on the replication feature. For more information, see WAN bandwidth requirements.
- What are your redundancy requirements? When implementing a replication solution, evaluate the redundancy needs for your system. For complete disaster recovery, it is important to ensure that each major component has a failover option. For the host server functionality, you can use cluster software with a separate physical or virtual backup. To protect data, use RAID mirroring in your storage solution. Also, using bonded network links to separate NICs protects network connections.
Hardware and software reference architecture describes the platform, hardware, and software requirements in detail.