HP Integrity rx1620 Server 1.30 GHz Windows Developers Bundle AB430A#007 プリント
製品コード
AB430A#007
The CLW architecture
The CLW architecture demonstrates a unique combination of these core technologies:
• Cluster parallel distribution
• Event-data-specific storage organization
• A non-transactional model
• Event-data-specific storage organization
• A non-transactional model
Cluster parallel distribution
The CLW leverages clustered processing node architecture to distribute workload and achieve parallel
The CLW leverages clustered processing node architecture to distribute workload and achieve parallel
computing on a massive scale.
Incremental scalability
It distributes workload equally among its clustered processing nodes enabling concurrent processing.
This allows the CLW to increase the throughput and usage of each node. Storage and throughput
It distributes workload equally among its clustered processing nodes enabling concurrent processing.
This allows the CLW to increase the throughput and usage of each node. Storage and throughput
capacity can be increased simply by adding more processing nodes.
The resulting increase in throughput and storage capacity is directly proportional to the number of
The resulting increase in throughput and storage capacity is directly proportional to the number of
processing nodes added. For example, adding a processing node to a three-node cluster would
increase storage and throughput capacity by 60 percent. Event data management capacity scales
increase storage and throughput capacity by 60 percent. Event data management capacity scales
directly with business growth. Cost and capacity planning are easily accomplished.
Distributed loading
As data is loaded into the CLW, it is evenly and concurrently distributed across all the processing
nodes in the cluster. This improves load efficiency.
As data is loaded into the CLW, it is evenly and concurrently distributed across all the processing
nodes in the cluster. This improves load efficiency.
Distributed search
Search requests received by the CLW are also evenly distributed on the processing nodes. Each node
Search requests received by the CLW are also evenly distributed on the processing nodes. Each node
conducts its portion of a search in parallel with the other nodes. The final results from each node are
aggregated and returned to the user.
Distributed aggregation
Aggregation searches that use ‘GROUP BY’ are distributed across the CLW processing nodes for
Aggregation searches that use ‘GROUP BY’ are distributed across the CLW processing nodes for
complete parallel processing of the aggregation operation. The results of each node are then
aggregated into a final result.
Data redundancy
Every event is recorded twice in the CLW cluster. Each copy is stored on a separate processing node.
Every event is recorded twice in the CLW cluster. Each copy is stored on a separate processing node.
A processing node that holds the copy of a failed node’s event data is designed to automatically take
over all search operations for the failed node. Searches and loading continues with the throughput
degraded by a factor proportional to the number of servers in the cluster. For example, in a five-node
cluster, throughput is reduced by 20 percent should a single processing node fail. This design enables
cluster, throughput is reduced by 20 percent should a single processing node fail. This design enables
the CLW architecture to provide high availability with marginal loss of performance.
Event data specific storage organization
The CLW storage organization is specific to the unique nature of event data, and produces significant
The CLW storage organization is specific to the unique nature of event data, and produces significant
advantages when managing the data.
13