Storage Networking Solutions for High Performance Databases

Storage Networking Solutions for High Performance Databases
Technology Brief
Storage Networking Solutions for
High Performance Databases
Optimized Performance and I/O Efficiency for Data
Warehouse Deployments
Industry Challenges
The exponential growth of data transactions and analysis is generated by a multitude of data sources. Business intelligence analytics
creates I/O-intensive requests on the data warehouse ecosystem. It has contributed to ever-increasing demands in the enterprise for
the following:
•• Higher bandwidth
•• Improved availability
•• Dynamic I/O provisioning
In addition to resolving the challenges presented by data growth, data warehouse systems require guaranteed Quality of Service (QoS)
for consistent, end-to-end performance to meet Service Level Agreements (SLAs), even under varying workload conditions.

Technology Brief
Table 1: Characteristics of a High-Performance I/O Interconnect Essential for
Complementing High-Performance Databases
METADATA LAYER
Technology Benefits
Extract, Transform,
Load (ETL)
Export to Excel, PDF
etc. for Offline Viewing
High I/O bandwidth for performance scale out
Executive
Dashboard
SOURCE DATA
LAYER
DATA TRANSFORMATION
LAYER
Faster Business
Analytics
Operational
& Analytical
Reports
Interactive Query
& Reporting
DATA WAREHOUSE
LAYER
•• More data warehouse transactions
•• Faster throughput and response
•• Faster backup and restore
REPORTING LAYER
High availability
OPERATIONS LAYER
Improve Service
Up Time
Figure 1. Data Flow
•• Prevent single point of failure
•• Improve I/O reliability for transactions
Data Warehouse Deployments
Reduce Infrastructure
Footprint
(Lower CAPEX)
Primarily used in analyzing queries for business intelligence, data
warehouse activities include the following: extract, transform, load (ETL),
backup and recovery, analytical processing, management services, and
traffic bursts. Large volumes of data sets are extracted and transported
for further data transformation and analysis. Data sources stationed in
singular or distributed locations across multiple nodes must be moved
into the data warehouse for subsequent analytical processing. ETL type
of processing with massive volumes of data in execution workloads rely
heavily on the CPU and I/O system resources. Typical data flow through the
data warehouse environment is shown in Figure 1.
Improve I/O bandwidth efficiency
•• Reduce underutilization of I/O resources
•• Avoid overutilization of I/O resources
Dynamic provisioning balances I/O workloads
•• Consistent end-to-end performance
Simplify System
Management, Tools,
and Methodology
(Lower OPEX)
•• Allocate I/O resources where required
•• Switch-agnostic
•• Flexible I/O allocation
•• Operating system (OS) agnostic
•• Flexible SAN and LAN functions
The I/O performance for the data warehouse is expected to be balanced
for providing end-to-end, consistent bandwidth. The elements of I/O
connectivity include Fibre Channel Host Bus Adapters and 10GbE
Converged Network Adapters that connect servers, switches, and
storage. The scalability of data warehouse workloads and I/O throughput
performance is typically linear and therefore scalability is essential for
managing end-to-end system resource allocation. Resource starvation
is observed when bottlenecks are created by unmet needs of increasing
workloads for dynamic throughput provisioning or additional I/Os to meet
network requests. Because the shared resources are limited, it is important
to maximize the efficiency of the system, improving CPU and I/O utilization
and thereby eliminating the bottlenecks. Table 1 is a checklist of the
characteristics of a high-performance, scalable I/O interconnect required to
complement a high-performance database transaction infrastructure.
TB12001
•• Single-pane-of-glass management
CPU efficiency saves system CPU cycles
Higher Application
Scale Out
•• Fewer CPU cycles used for I/O
•• More room for application scalability
•• Run more data warehouse operations
QoS provides guaranteed I/O provisioning
Meet SLA
Compliance
•• Minimum and maximum metering
•• Fine grain QoS
Meet Unexpected
Demands
Oversubscription leverages unused
bandwidth reserve
•• Manage unexpected I/O usage spikes
•• Use same hardware for additional I/O
SN0330974-00 Rev. B 02/13

2
Technology Brief
QLogic Solution
I/O bottlenecks are removed by providing sufficient bandwidth when and
where it is required. Oversubscription can be used for meeting unexpected
I/O workload spikes beyond the minimum guaranteed bandwidth.
Data center consolidation driven by server virtualization and storage
network convergence is accelerating the transition to 10GbE technology
to complement the needs of high throughput and I/O-intensive database
deployments. ConvergeFlex® and VMflex™ solutions from QLogic®
support QoS to extend the benefits of virtualization by reducing TCO on
cables, ports, PCIe® slots and improving CPU and I/O utilization efficiency
for end-to-end system performance. High-performance data warehouse
systems require I/O throughput optimization (higher bandwidth), system
CPU efficiency (improved availability), and system I/O utilization efficiency
(dynamic provisioning). All of these factors are critical for meeting industry
challenges and achieving business goals.
Numerous operations in the data warehouse ecosystem are executed
in parallel, such as accessing and processing large volumes of data for
analytics, updates, inserts, and deletes. The performance of I/O-intensive
operations relies on the overall efficiency of the system’s performance
and the availability of guaranteed scalable bandwidth. QLogic’s solution
helps mitigate the overutilization and underutilization of resources that
can otherwise lead to general system performance degradation. QLogic
provides a simple, dynamically configurable, high availability solution with
scalable bandwidth to meet the I/O performance and efficiency demands of
data warehouse deployments.
The QLogic solution, designed for high I/O throughput performance and
fine grain QoS with dynamic bandwidth provisioning, is well suited for
data warehouse deployments. Primarily used in analyzing queries for
business intelligence, data warehouse activities—extract, transform, load
(ETL), backup and recovery, analytical processing, management services,
and traffic bursts—use large volumes of data sets. The data sets are
extracted and transported using high performance and network bandwidth.
Data sources that are stationed in singular or distributed locations across
multiple nodes are moved into the data warehouse for subsequent data
transformation and analytical processing. Executing ETL processing with
massive volumes of data in parallel execution creates significant demand
on the CPU and I/O system resources. Data warehouse transactions
typically are characterized by the following model:
Guaranteed
Provisioning
Mgmt.
DR
Log Files
Database
Transactions
Data
Warehousing
Hypervisor
Data Mart
Database
Clients
Data Mining
Remote
Office
Mobile Access
QLogic
8200 Series
Converged Network
Adapters
NIC
NIC
NIC
FCoE
Cloud
Storage
REMOTE SAN
(off-site)
LAN
QLogic
ISR6200
Data
Access
Server
Executive Dashboard
Business
Intelligence
SAN
ENTERPRISE DATABASE
WAREHOUSE
QLogic 5800 Series
Switch with FCF
QLogic
2500 Series
Fibre Channel
Adapters
FC Storage
KEY ATTRIBUTES OF QLOGIC SOLUTION
Operational & Analytical
Reports
Interactive Query Export to Excel, PDF
& Reporting
etc. for Offline Viewing
• 10Gb Solutions for High Bandwidth
• Flexible Personality for Improved Availability in LAN and SANs
• QoS for Dynamic Provisioning
10GbE Switch
NAS
Storage
QLogic 3200
Intelligent Ethernet
Adapters
Figure 2. The QLogic Solution in the Database Warehouse
•• Sequential I/O access
Related Video: QLogic Adapter of Choice for Database
Performance
•• Large I/O transactions
•• I/O size range: 64k to 1MB
•• DB block size: 8KB (or larger)
•• Parallel execution for scaling (if not limited by CPU or I/O)
Efficient I/O utilization in the hardware is achieved by allocating the
precise amount of bandwidth resources to the applications running on
the virtualized or native host. The technology significantly improves
resource availability, reduces the possibility of I/O underutilization, and
prevents wasted bandwidth. An administrator can change the I/O function’s
personality between storage or networking on the same hardware. The
10GbE converged solutions provide the best storage and networking
capabilities to match the sustained throughput requirements. Large block
and I/O transactions take up a bigger slice of the bandwidth. The I/O offload technology (SAN or LAN) frees up the system CPU cycles, especially
during parallel execution of the workloads when the demand on the system
CPU is further stressed to process the data warehouse transactions. Fine
grain QoS allows minimum guaranteed bandwidth allocation and maximum
metering. The resource allocation can be dynamically managed to align
and respond to varying degrees of bandwidth workload conditions. The
TB12001
Click on the Image to Watch a YouTube Video
SN0330974-00 Rev. B 02/13

3
Technology Brief
Follow us:
Share:
Corporate Headquarters
www.qlogic.com
International Offices
QLogic Corporation
26650 Aliso Viejo Parkway
Aliso Viejo, CA 92656
949-389-6000
UK | Ireland | Germany | France | India | Japan | China | Hong Kong | Singapore | Taiwan
© 2013 QLogic Corporation. Specifications are subject to change without notice. All rights reserved worldwide. QLogic, the QLogic logo, ConvergeFlex, and VMflex are trademarks or registered trademarks of QLogic Corporation. PCIe is a registered trademark of PCI-SIG Corporation.
All other brand and product names are trademarks or registered trademarks of their respective owners. Information supplied by QLogic Corporation is believed to be accurate and reliable. QLogic Corporation assumes no responsibility for any errors in this brochure. QLogic Corporation
reserves the right, without notice, to make changes in product design or specifications.
TB12001
SN0330974-00 Rev. B 02/13

4
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertising