Social Icons

Scalabilty

"Scalability" means that as you add components to the system, the performance increases continuously.
That means if you double your AMPs you will double your speed. A 4-AMP system can double its speed by adding 4 more AMPs to become an 8-AMP system.Its enables the system to support more users/data/queries/complexity of queries without facing performance issues.

In the picture on the following page you can see we have a 4,000 AMP system
TeradataWiki-Teradata Scalability


The chart below lists the meaning of the prefixes:

Teradata Scalabilty

The Teradata Database can scale from 100 gigabytes to over 100+ petabytes of data on a single system without losing any performance capability. The Teradata Database's scalability provides investment protection for customer's growth and application development. The Teradata Database is the only database that is predictably scalable in multiple dimensions, and this extends to data loading with the use of parallel loading utilities. The Teradata Database provides automatic data distribution and no reorganizations of data are needed.

2 comments:

  1. Although the scalability sounds great in theory, keep in mind that parallel databases like Teradata offer no high level of fault tolerance with increasing number of nodes.

    Roland (http://www.teradatapro.com).

    ReplyDelete
  2. There are multiple levels of fault tolerance and high availability built into the Teradata Database system with the best features allowing customers to run 24x7.

    At the lowest level are the hardware system components:
    • RAID 1 mirroring to protect against disk failure.
    • Hot pluggable disk drives (Processing Node Chassis)
    • Hot pluggable power supplies (Processing Node Chassis, 3GSM Chassis)
    • Hot pluggable Distributed Management Boards (3GSM Chassis, Processing Node Chassis)
    • Redundant fans (Processing Node Chassis, 3GSM Chassis, BYNET Chassis)
    • Redundant power supplies (Processing Node Chassis, 3GSM Chassis, BYNET Chassis)
    • Redundant Interconnect (BYNET Chassis)
    • Redundant I/O buses (Processing Node Chassis)
    • Parity checking on I/O busses (Processing Node Chassis)
    • EDAC (Error Detection and Correction) on memory (Processing Node Chassis)
    • Flash EEPROM update (Processing Node Chassis, 3GSM Chassis, BYNET Chassis)
    • Field Serviceable Media Drives, Processors, Memory, I/O (Processing Node Chassis)
    • Chassis Field Replaceable Units (All Chassis)
    • Remote support capability (All Chassis)
    • Dual AC Power (All Chassis and cabinets)
    • Battery backup for dumping critical data to disks upon power fail recovery (UPS Chassis, 3GSM Chassis, Processing Node Chassis)
    • Internal UPS (UPS Chassis)
    • Environmental monitoring; i.e., ,temperature, power
    At the next level, the Teradata SMP nodes are grouped together in cliques (two to three node clusters) where every node is interconnected with every other node’s disk drives (Figure 9). This is an out-of-the-box feature that every Teradata installation has. When a node fails within a clique, the AMPs that were running on the failed node will distribute to the remaining clique nodes after a system restart (which is normally three minutes) enabling work on the system to resume albeit in a reduced performance mode (33 percent reduction in a three node clique).

    An additional availability feature available to customers is a hot standby node. This feature works with any clique size. The hot standby node is fully configured with all the necessary software and is ready to go. When a node fails, all of the amps on the failed node transfer to the hot standby node after a system restart (which is normally three minutes). There is no performance impact during this node outage scenario. After the failed node is repaired, it is designated as the hot standby node and operations continue uninterrupted without another system restart.

    An additional availability feature available to customers is a hot standby node. This feature works with any clique size. The hot standby node is fully configured with all the necessary software and is ready to go. When a node fails, all of the amps on the failed node transfer to the hot standby node after a system restart (which is normally three minutes). There is no performance impact during this node outage scenario. After the failed node is repaired, it is designated as the hot standby node and operations continue uninterrupted without another system restart.

    The highest level of availability is for those customers that cannot afford to have their system go down during specific periods of operation. This feature is called Dual Active; where two separate Teradata systems are synchronized and either system can service a data request from the users. This feature also serves as a disaster recovery solution when the two systems are geographically separated.
    The advantages of this feature include load balancing workloads across the two systems to maximize performance throughput, and if one system fails then the other system will assume all the workload. This is also useful for scheduled system maintenance. The two systems can stagger the service and still complete the work coming from the data warehouse users

    ReplyDelete