Vous êtes sur la page 1sur 9

Achieving Oracle Database

Scalability using SharePlex


Complement or replace RAC with enterprise-class database replication technology.

Written by Sujith Kumar, Chief Technologist, Dell Software

Introduction servers. Specifically, RAC supports seamless deployment of one


Users demand high performance, but databases are or more servers accessing the same database within the same
constrained by hardware limits. cluster, providing fault tolerance, high availability and horizontal
Due to the so-called “Google effect,” users have come to scalability within the same data center. However, RAC has
expect rapid response from applications, and that demand for important drawbacks, including platform options, geographical
performance trickles down to the underlying databases. But availability and, most importantly, cost.
the hardware supporting traditional relational databases has a
finite limit in terms of system resources: hardware resources SharePlex delivers cost-effective database replication that
such as CPU, memory, network and storage have a vertical overcomes RAC’s shortcomings.
scalability limit. This tech brief offers a better option: SharePlex from Dell
Software. SharePlex is an enterprise-class logical database
Oracle RAC enables IT organizations to scale horizontally replication technology that allows data to be replicated
using cheaper hardware—but it is costly. between databases in the same data center or across data
To help, Oracle introduced Oracle Parallel Server (OPS) in centers with no distance limitations. With SharePlex, IT
Oracle 7, and then morphed OPS into Real Application Cluster organizations can overcome RAC’s limitations and achieve
(RAC) in Oracle 9i. Oracle RAC enables IT organizations to “six nines” of availability for their databases with no platform,
scale horizontally using cheaper hardware, making it a cost- geographic, network or performance limitations.
effective alternative to expensive single monolithic super
This tech brief is aimed towards Oracle • Load balancing–Since there is more than
DBAs, disaster recovery and high one node within a cluster, workloads can
availability (DR/HA) architects, and be distributed across multiple servers,
application architecture architects thereby optimizing system resources.
responsible for engineering, creating • Consolidation–The shared-everything
and managing application infrastructure architecture of RAC allows for
stack interested in achieving highly consolidation of disparate databases within
available and scalable Oracle databases the same cluster, enabling better utilization
focusing on data replication needs at and scheduling of system resources based
the same time. It assumes readers are on the needs of the different databases.
familiar with Oracle database technology
and Oracle RAC and that they have Drawbacks of RAC
a rudimentary understanding of data • Single point of failure (SPOF)—Oracle
RAC requires replication options. RAC’s shared-everything architecture
makes the database a single point of failure.
specialized and Why consider RAC? If the RAC experiences a storage failure,

expensive hardware, Traditionally, the only option available it will stop working and cause the whole
to IT organizations facing increasing database to fail. Moreover, all the instances
not only for the database size and load was to scale and the database within a given RAC are
vertically. Vertical scaling involved orchestrated by Oracle Clusterware, which
servers but also for investing in expensive proprietary constitutes another SPOF.

the network and hardware. With the release of Oracle • Downtime during upgrades—Since all
9i in 2001, Oracle unveiled the re- the instances in RAC are accessing the
storage layers. invigorated and re-designed version same database, downtime is required for
of Oracle Parallel Server (OPS) called upgrades to the database server (such
Oracle Real Application Cluster. Oracle as hardware changes, operating system
RAC enables transparent and concurrent upgrades, Oracle upgrades, storage
access to a single database by multiple replacement and application upgrades).
servers, each with its own independent The rolling upgrades features in Oracle 10g,
instance, thereby providing fault 11g and 12c support only minor version
tolerance, high availability and affordable upgrades of the Oracle binaries in
horizontal scalability. RAC is included RAC instances.
as part of the Oracle Standard Edition • Geographic limitations—All the nodes
database license, and RAC is considered in a cluster must be hosted in the same
a paid add-on with Oracle Enterprise data center. This is a requirement since
Edition database. all the nodes will have to be connected
to each other using expensive network
Benefits of RAC interconnects such as GigE or Infiniband
• Scalability—RAC allows seamless horizontal network interfaces. RAC supports extended
scaling by enabling transparent deployment distance clusters (that is, clusters across
of servers into a cluster that supports user data centers); however, the maximum
and data growth. distance between data centers is 100
• Availability–Since multiple instances kilometers. For longer distances, Oracle
running on different servers are accessing recommends using Oracle Data Guard
the same database, the database itself is between the source and target data centers.
impervious to server and network failures; • Performance limitations—RAC allows
the user workload gets shifted from the the same workload distribution across
failing or failed node to one of the other multiple nodes within the cluster. However,
surviving nodes. This is accomplished because all the instances are accessing
using RAC technologies such as Single the same database, there is no way to tune
Client Access Name (SCAN), Transparent each instance or node for a different type
Application Failover (TAF) and Failover of workload, such as online transaction
Application Notification (FAN). processing (OLTP), data mining or reporting.

Share:
2
• Platform limitations—RAC does not placed SharePlex as the #1 third-party
support heterogeneous hardware within database replication solution, with over
the same cluster, such as a mixed platform 19 percent of the database replication
of a combination of HP-UX, Oracle Solaris, market compared to Oracle’s native
IBM AIX or Linux x86 servers. A cluster also tools at 25.5 percent (inclusive of
cannot span multiple versions of Oracle, GoldenGate, Data Guard, Active Data
such as 10g, 11g and 12c. Guard and Oracle Streams).
• Costs—Due to its shared-everything
architecture, RAC requires specialized What is SharePlex?
and expensive hardware, not only for SharePlex is an affordable, easy-to-use
the servers but also for the network and logical data replication solution for
storage layers. It also requires an additional Oracle databases. It provides a real-
highly skilled IT professional to implement, time copy of production data without
manage and maintain it. Furthermore, impacting the primary database’s SharePlex from
Oracle charges an additional licensing fee performance and availability.
for the RAC add-on when it is used with Dell Software is an
Oracle Enterprise Edition database. The data is replicated from the source affordable logical
database to one or more target database
The cost-effective servers. SharePlex runs continuously replication solution
alternative: SharePlex on the database server itself, replicating
SharePlex from Dell Software is an data in real time with sub-second
that can either
affordable logical replication solution
that can either complement Oracle
latency and no perceivable impact on complement Oracle
the source database performance. On
RAC or completely replace it. Easy to the source database, the transactional RAC or completely
use and manage, SharePlex can help data changes are captured directly from
IT organizations achieve horizontal the Oracle redo logs or archive log files
replace it.
scalability by providing one or more using a proprietary capture mechanism.
real-time copies of the production RAC Applications are able to access multiple
or single-instance databases with no copies of the database seamlessly—an
distance limitations. analogue to receiving e-mails in desktop
Microsoft Outlook client and mobile
SharePlex has been a leading devices simultaneously. SharePlex can
database replication product since replicate the database in its entirety
1998 and boasts a global customer or can be limited to specific schemas,
base. In its November 2012 report, tables, columns or rows within tables.
Worldwide Database Development and It also has the ability to modify or
Management Tools 2011 Vendor and transform replicated data on the fly prior
Segment Analysis (Doc #237963), IDC to posting it to the target database.

Export Import
Export Post
queue queue

SQL

Read
Post Post
Redo/archive logs

Capture Capture
queue
Source Cloud target Target
Figure 1. SharePlex captures changes to the source database and keeps the replica in
sync in real time.

Share:
3
• Optimization for different workloads–
How does SharePlex work?
RAC allows for the same workload to be
SharePlex operates on both source
distributed across various nodes operating
and target database servers. It spawns
within the same cluster and does not allow
capture, read and export processes
disparate workloads to be run on different
on the source database server, and
nodes. SharePlex maintains multiple copies
import and post processes on the
of the same database in real time, so each
target database server. The capture
replica database can be customized for
process scans the Oracle redo logs for
different workloads. For example, even
any changes to the data and captures
if the main source database is optimized
the changes. If necessary, the capture
for online discrete transactions, you can
process has the ability to scan the
optimize a replica for reporting by creating
archive logs instead of the online redo
a different set of indices that will run
logs. The read process prepares the data
With RAC, if the to be transported across the network

reports and queries more efficiently.
Geographic flexibility–As noted earlier,
after processing it and hands it over
database goes the export process. The export process
RAC cannot span data centers over long
distances. SharePlex, on the other hand,
down, the entire transports the data to one or more
target database servers via TCP/IP across
uses TCP/IP to communicate between

cluster will fail. the LAN or WAN.


data centers via LAN/WAN with no distance
limitations, thereby reducing the total cost
SharePlex eliminates On the target database server(s), the
of implementation.
• Heterogeneous platform & version
this single point import process intercepts the TCP/IP
packets and process the data received
support–Oracle RAC requires

of failure by from the source database server and


homogeneous hardware architecture
and all nodes must run the same version
hands it over to the post process. The
creating one or post process constructs SQL statements
of Oracle. Since SharePlex is agnostic
with respect to operating system, CPU
more replicas of for the DML operations based on
the data changes observed on the
architecture and database version, it can

the database; if source database and executes them by


support any mix of operating system and
database versions, including Oracle 9i to
connecting to the target database(s).
the database goes Oracle 12c. Replication between Oracle
Enterprise Edition and Oracle Standard
down, you can SharePlex also has the ability to encrypt
and compress the data while being
Edition is also supported.

simply flip the users transported across the network to


• ZeroIMPACT upgrades & migrations–As
we saw, rolling upgrades in RAC are limited,
enhance security and performance. High
or the application throughput and resiliency is achieved by
and schema object changes, network
or storage infrastructure upgrades, and
to a copy of the transmitting the data in memory under
ideal circumstances. Organizations that
application upgrades require downtime.

database with no need fault tolerance and high availability


Since SharePlex maintains two or more
distinct copies of the same database, it
can persist the data in proprietary disk-
downtime. based queue files.
can fully support all types of upgrades
and migrations with near zero downtime.
You can upgrade the operating system,
Benefits of SharePlex
storage and network infrastructure, servers,
SharePlex offers some of the same
databases and applications with no impact
benefits as Oracle RAC while eliminating
to your applications or users.
its drawbacks:
• Database consolidation–A single
• No single point of failure (SPOF)–As
implementation of RAC allows multiple
we saw, with RAC, if the database goes
databases to run on the same node, sharing
down, the entire cluster will fail. SharePlex
the same storage, so if the storage fails,
eliminates this single point of failure by
all the databases will become unavailable.
creating one or more replicas of the
SharePlex allows you to consolidate
database; if the database goes down, you
databases on the same node using different
can simply flip the users or the application
storage mount points. This eliminates
to a copy of the database with no downtime.

Share:
4
the single point of failure: even if one of failover technologies, such as FAN and
the storage mount points fails, the other TAF, without re-writing application logic,
databases and the associated instances unlike RAC, which requires application
running on a given node will continue to modification
function. Moreover, since the database can
be replicated to more than one node at the What about Oracle Standard Edition?
same time, total failures are prevented. Small to medium size companies often
• Reduced costs–As noted above, RAC implement RAC as part of Oracle
requires specialized and expensive Standard Edition. They therefore face
hardware. SharePlex, on the other hand, certain limitation, including the number
can run on low-cost or re-purposed of CPUs in a server, that SharePlex can
hardware, allowing it scale horizontally help them overcome.
at an affordable cost. You can add new
commodity servers without incurring SharePlex will also help create low-cost In IT organizations
the cost of replacing the existing more Oracle Standard Edition deployments
expensive and older servers, so you can that can be used for development and where Oracle

scale up more cost effectively.
Load balancing– Since all the nodes within
testing without incurring additional
cost overhead.
RAC has been
a RAC access the same physical databases, implemented,
even though the user load can be spread How organizations are using
across multiple nodes, the users access the SharePlex today SharePlex can
same database, which will end up being a
bottleneck at some point. With SharePlex,
Overview
Organizations use SharePlex for a wide
overcome some of
the database and its replica(s) can be set up range of purposes. Here are a few the shortcomings
in multi-master or bi-directional replication examples on how some customers use
mode, so you can spread the user load SharePlex in a RAC environment. of RAC at a much
across multiple databases and also across
multiple data centers. Case study 1: Dell IT: Siebel 10g to 11g
lower cost.
database migration
How SharePlex complements RAC Dell needed to migrate a 10TB
In IT organizations where Oracle RAC business-critical database running a
has been implemented and is an integral Siebel application from Oracle 10g
part of the enterprise, SharePlex can RAC to 11g RAC. The application
overcome some of the shortcomings of served both internal users and 22,000
RAC at a much lower cost. By providing customers worldwide, so it required
one or more copies of the primary 99.99% uptime—only 4.38 hours of
database on separate servers within annual downtime were available for
the same data center or across data maintenance and upgrades.
centers, SharePlex:
• Provides a shared-nothing architecture To upgrade the database while
that eliminates the single points of failure minimizing downtime, IT staff followed
discussed above this procedure:
• Supports replication to and from a virtually 1. RMAN was used to perform the initial copy
unlimited number of nodes using the of the 10g database from the old hardware
cascading configuration, overcoming to the new hardware. Since Linux was the
Oracle 11gR2’s limitation restricting each operating system on both the old and new
RAC implementation to a maximum of hardware platforms, the OS was not
100 nodes a concern.
• Eliminates geographic limitations by 2. SharePlex was installed on both source
removing RAC’s distance limits and target servers and was configured to
• Supports multiple operating system types, capture data on the source while the post
CPU architectures (x86, x64, SPARC and or apply process on the target server was
RISC) and database versions stopped and the transactional data was
• Enables you to take advantage of new allowed to queue up on the target server.

Share:
5
10g, 4-node RAC, 11g, 5-node RAC,
RedHat Linux 4, RedHat Linux 5,
Siebel application Siebel application
Figure 2. SharePlex migration with reverse replication for failback, if required

3. Oracle Data Guard was established Reverse replication from Oracle 11g to
One team was able between the source 10g database and
the target 10g database running on the
10g was configured as a failback option,
in case it would be required.
to reclaim 1.2TB of new hardware platform. Once Data Guard The Dell DBA team summarized the
caught up with the logs, it was stopped and benefits of using SharePlex for the
storage space in an the database was recovered. At the time migration as follows:

11g environment of database recovery, the SCN number of


the last transaction was noted and the 10g
• SharePlex allowed the migration and
upgrade to be completed six weeks ahead
by performing database on the new server was upgraded of schedule, and allowed enough time for
to 11gR2. extensive testing.
the archive/ 4. The transactions in the SharePlex queue • The team was able to reclaim 1.2TB of

purge during the were reconciled (duplicate transactions


were eliminated from the queue) to the
storage space in the 11g environment by
performing archive/purge. They paused
migration. SCN noted in the earlier step, and the SharePlex during the archive/purge
remaining data was applied to the process and restarted it after the process
new database. was complete.
5. Once the SharePlex queue was empty, the • The project resulted in an improved
10g and the 11g database were considered user experience.
to be in sync and users were switched over • SharePlex enabled the team to test multiple
to the new 11g database. environments in the new platform with
no downtime.

Figure 3. Using SharePlex active-active replication across multiple data centers

Share:
6
RAC 1: Airline reservations RAC 2: Hotel reservations

Figure 4. SharePlex peer-to-peer replication using a RAC configuration

Case study 2: Horizontal scaling with In addition to providing the HA/DR


HA & DR options
A Dell customer has multiple data
functionality, SharePlex replication also
enabled application, OS and Oracle
Using SharePlex
centers spread across the U.S., with version upgrades with near-zero in active/active
multiple RAC implementations at each downtime.
data center. However, the limitation of replication
RAC make it impractical, if not impossible,
to link the RAC implementation in each
Case study 3: Peer-to-peer replication
using RAC
configuration mode,
data center either to work in tandem to A well-known e-commerce company the customer was
service the user load or to be configured uses SharePlex to replicate data between
in HA/DR mode. two RAC databases residing in two able not only to
Using SharePlex in active/active
different data centers. One of the RAC
environments has two databases for
scale horizontally
replication configuration mode, the processing transactions and queries within the same
customer was able not only to scale from booking agents, direct channels
horizontally within the same data center and online customers supporting the data center but also
but also to have each data center act as
HA/DR for the others. The configuration
first product offering. The second RAC
environment has another two databases
to have each data
of data replication between multiple for handling transactions and queries center act as HA/DR
data centers using SharePlex is shown from booking agents, direct channels
in Figure 4. and online customers supporting a for the others.
second product offering.

Data center 1 Data center 2

Target 1: Billing Target 2: Reporting


(internal users) (external users)
Figure 5. RAC replicating to reporting single instance

Share:
7
Using SharePlex peer-to-peer replication, Conclusion
the customer has been able to offload It’s inevitable that your user load and data
querying and report processing to the size will grow along with your business.
least loaded environments without At some point, vertical scaling of your
impacting the user experience on databases will become cost prohibitive
the source database. In addition, the or physically impossible to attain.
customer can perform customized Oracle’s answer to this problem is Oracle
database backups, upgrades and RAC. However, RAC has important
maintenance activities without impacting disadvantages, including high hardware
users, and site outages due to database and support costs, a single point of
malfunction have been eliminated. failure, downtime for upgrades and
migrations, and limited ability to tune for
Case study 4: Replication from Oracle performance based on user workloads.
SharePlex can RAC to Oracle single instance
An online collaboration company with SharePlex can be used either in
be used either a data center in San Jose, California, conjunction with RAC or as a
in conjunction and another in Denver, Colorado, has
implemented multiple RACs in both data
replacement for RAC to provide
seamless horizontal scaling within the
with RAC or as a centers to handle online meetings by same or across data centers without the
users. Each data center acts as a HA/DR costs or other drawbacks associated
replacement for RAC site for the other. Both data centers are with RAC.
to provide seamless considered to be active, since they are
used for load balancing of users as well. About the author
horizontal scaling Sujith has more than 20 years of
SharePlex is used to replicate data experience in a variety of database
within the same or between the RAC environments in the technologies. For the past 15 years,
across data centers two data centers in a bi-directional
configuration. In addition, SharePlex
he has been working on SharePlex,
architecting custom data management
without the costs replicates selected data from both data solutions for Fortune 500 companies.
centers to two separate single instance As one of Dell’s chief technologists, he
or other drawbacks databases, which are optimized for has been involved in acquisitions and
associated with RAC. customer billing and external reporting. mergers of other software companies.
Sujith holds a M.S. in Engineering from
Using SharePlex in this configuration Texas A&M University and a B.S. in
has enabled this customer to achieve Engineering from Bangalore University.
unparalleled high availability that Oracle
RAC cannot provide, leading to a better
user experience and enhanced
customer satisfaction.

Share:
8
For More Information
© 2014 Dell, Inc. ALL RIGHTS RESERVED. This document DELL ASSUMES NO LIABILITY WHATSOEVER AND DISCLAIMS
contains proprietary information protected by copyright. No ANY EXPRESS, IMPLIED OR STATUTORY WARRANTY RELATING
part of this document may be reproduced or transmitted in TO ITS PRODUCTS INCLUDING, BUT NOT LIMITED TO, THE
any form or by any means, electronic or mechanical, including IMPLIED WARRANTY OF MERCHANTABILITY, FITNESS FOR
photocopying and recording for any purpose without the A PARTICULAR PURPOSE, OR NON-INFRINGEMENT. IN NO
written permission of Dell, Inc. (“Dell”). EVENT SHALL DELL BE LIABLE FOR ANY DIRECT, INDIRECT,
CONSEQUENTIAL, PUNITIVE, SPECIAL OR INCIDENTAL
Dell, Dell Software, the Dell Software logo and products—as DAMAGES (INCLUDING, WITHOUT LIMITATION, DAMAGES
identified in this document—are registered trademarks of Dell, FOR LOSS OF PROFITS, BUSINESS INTERRUPTION OR LOSS
Inc. in the U.S.A. and/or other countries. All other trademarks OF INFORMATION) ARISING OUT OF THE USE OR INABILITY
and registered trademarks are property of their respective TO USE THIS DOCUMENT, EVEN IF DELL HAS BEEN ADVISED
owners. OF THE POSSIBILITY OF SUCH DAMAGES. Dell makes no
representations or warranties with respect to the accuracy or
The information in this document is provided in connection completeness of the contents of this document and reserves
with Dell products. No license, express or implied, by estoppel the right to make changes to specifications and product
or otherwise, to any intellectual property right is granted by descriptions at any time without notice. Dell does not make
this document or in connection with the sale of Dell products. any commitment to update the information contained in this
EXCEPT AS SET FORTH IN DELL’S TERMS AND CONDITIONS AS document.
SPECIFIED IN THE LICENSE AGREEMENT FOR THIS PRODUCT,

About Dell
Dell Inc. (NASDAQ: DELL) listens to customers and delivers
worldwide innovative technology, business solutions and
services they trust and value. For more information,
visit www.dell.com.

If you have any questions regarding your potential use of


this material, contact:

Dell Software
5 Polaris Way
Aliso Viejo, CA 92656
www.dell.com
Refer to our Web site for regional and international
office information.

Share:
9

TechBrief-SharePlex-OracleDB-US-KS-24265

Vous aimerez peut-être aussi