Welcome to Scribd, the world's digital library. Read, publish, and share books and documents. See more
Standard view
Full view
of .
Look up keyword
Like this
0 of .
Results for:
No results containing your search query
P. 1
Exploração Das Funcionalidades de Alta Disponibilidade No Microsoft SQL Server 2005

Exploração Das Funcionalidades de Alta Disponibilidade No Microsoft SQL Server 2005

Ratings: (0)|Views: 41|Likes:

More info:

Published by: Paulo Henrique G Fagundes on Jan 17, 2008
Copyright:Attribution Non-commercial


Read on Scribd mobile: iPhone, iPad and Android.
download as PDF, TXT or read online from Scribd
See more
See less





Reprinted from
Dell Power Solutions,
February 2006. Copyright © 2006 Dell Inc. All rights reserved
February 2006
terprise data centers that host business-critical appli-cations and databases can be at risk from failures anddisasters. Availability and recoverability become crucial forbusiness operations. Current database availability technolo-gies vary in their complexity, cost, level of automation,incurred downtime, and supported distance. A major chal-lenge for an enterprise is determining the most cost-effectiveand least complex solution to help ensure database avail-ability and meet business needs. Microsoft SQL Server 2005introduces several high-availability features and featureenhancements for the SQL Server database platform.
Failover clustering
Failover clustering with Microsoft Cluster Service (MSCS)is a popular way to achieve high availability for SQLServer. With MSCS, multiple servers or nodes are linkedto function as a single system and provide an automaticfailover solution. Enterprises are connected to virtual serv-ers and not physical servers; each active physical servermay host one or more virtual servers.If one of the physical servers fails because of either ahardware or software problem, MSCS detects the failureand moves resources that reside on the failing server—including the virtual server(s)—to one or more remaininghysical servers. End users connected to the failed serverobserve only a momentary delay in accessing its resourceswhile MSCS restarts SQL Server services on the remainingserver(s) and remaps the virtual server connections.Installing a failover cluster requires the following:
Microsoft Windows
2000 Server (Advanced Serveror Datacenter Server) or Microsoft Windows Server
2003 (Enterprise Edition, Datacenter Edition,Enterprise x64 Edition, or Datacenter x64 Edition)
Microsoft SQL Server 2005 (Standard Edition orEnterprise Edition)
Shared storage based on SCSI, Fibre Channel, orSerial Attached SCSI
All components (server, host bus adapter, and stor-age system) and the cluster solution listed in theMicrosoft Windows Server CatalogFigure 1 shows two typical two-node failover clusterconfigurations for SQL Server 2005, one in active/passive
Exploring High-AvailabilityFeatures
in Microsoft SQL Server 2005
SQL Server
2005 offers failover clustering, database mirroring, logshipping, replication, and database snapshot features that can help safeguard anenterprise’s critical data and operations. This article discusses these features andtheir associated trade-offs.
Related Categories: 
Clustering Database High availability (HA) Microsoft SQL Server 2005 
Visit www.dell.com/powersolutions for the complete category index.
www.dell.com/powersolutionsReprinted from
Dell Power Solutions,
February 2006. Copyright © 2006 Dell Inc. All rights reserved
mode and one in active/active mode. In a failover cluster envi-onment, SQL Server 2005 offers several additional features andmprovements compared to SQL Server 2000:
Support for multinode clustering:
Standard Edition allowsup to two-node clustering and Enterprise Edition up to eight-node clustering.
Support for SQL services:
Analysis Services feature offersmulti-instance support and full-text indexing.
Easier installation:
System Configuration Check automati-cally detects and installs SQL Server as virtual servers acrossthe nodes in the cluster.
Database mirroring
Database mirroring is designed to augment the availability of SQLServer 2005. With this feature, transactions to a database hostedon a SQL Server instance, referred to as the principal, can be con-inuously duplicated in real time onto a copy hosted on anothernstance, referred to as the mirror. The principal and mirror serversare considered partners in a database mirroring session. Transac-ions (such as insertions, updates, and deletions) on the principaldatabase are duplicated by continuously sending transaction logecords to the mirror over the network (see Figure 2).The mirror receives the log records continuously and restoreshem. If the principal or mirror server incurs a failure, the mirror-ng session is disconnected. Before starting the mirroring session,administrators must initialize the mirror database from a principaldatabase full-restore with the NORECOVERY option.Besides the primary and mirror, a mirroring session may includean optional third server, referred to as the witness. The witnessserver enables automatic failover after a principal server failure bypromoting the mirror to function as the principal. Using a witnessserver helps achieve a quorum and prevents accidental promotionof the mirror database that results from communication failuresbetween the principal and mirror.Database mirroring requires the following configuration:
he primary and mirror databases must be hosted on sepa-ate SQL Server 2005 instances.
he principal database must be set to the Full Recovery model,ot the Bulk-Logged Recovery or Simple Recovery model.Database mirroring can be configured in any of the followingthree operating modes based on transaction safety level and thepresence of a witness server:
High-availability mode:
In this mode, log records are trans-erred synchronously from the principal database to theirror. The principal waits for an acknowledgment from theirror before “hardening” log records to its disk. In additionto providing synchronous transfer, this mode uses a witnessserver to enable automatic failover.
High-protection mode:
This mode is similar to the high-availability mode except that it does not support a witnessserver, and thus automatic failover is not possible. However,failover can be performed manually to promote the mirrordatabase. Because the log records are transferred synchro-ously in this mode, the mirror database is synchronous andconsistent with the principal.
High-performance mode:
In this mode, log records aretransferred asynchronously from the principal to the mirrordatabase. The principal does not wait for an acknowledg-ent from the mirror before “hardening” its logs to the disk,
Figure 1. Failover cluster configurations for SQL Server 2005
SQL Server 2005db1db2SQL Server 2005HeartbeatLAN
SQL Server 2005db1SQL Server 2005HeartbeatLAN
Figure 2. Database mirroring for database db1 in high-availability mode
db1db2db3SQL Server 2005db5db6SQL Server 2005SQL Server 2005db1
Transaction log records Log recovery 
Reprinted from
Dell Power Solutions,
February 2006. Copyright © 2006 Dell Inc. All rights reserved
February 2006
and the mirror is not guaranteed to be synchronous with theprincipal at any point. This mode does not support a wit-ness server, and thus neither automatic nor manual failoveris possible. However, a forced failover can be performed torecover the mirror, with some data loss occurring becausemirroring is asynchronous.A key advantage of database mirroring is that it providesquick failover and high availability with minimal hardware costand complexity. Formerly, such SQL Server mirroring solutionswere often feasible only with expensive, proprietary hardware-based mechanisms. Withacceptable latencies andbandwidth, mirroring canbe deployed over a long-distance wide area network(WAN) as part of a disasterecovery solution.During the mirroringsession, the mirror data-base is not available forclient access because its in a recovery state, buta database snapshot canbe created on the mirrorfor limited reporting. Onlyone mirror database canbe established for a princi-al database. However, theparticipating SQL Servernstances can assume differ-ent roles for different mirror-ng sessions. For example,a SQL Server instance canserve as a principal for one database, as a mirror for anotherdatabase, and as a witness for yet another mirroring session. Thenotion of a virtual server, as is used in failover clustering, doesnot exist. However, Microsoft ADO.NET provides an applicationprogramming interface (API) for clients to automatically reconnecto the mirror server during failures.
Log shipping
Log shipping is similar to database mirroring. It allows a databaseosted on a primary server to automatically send transactionogs to a secondary database for duplication. The log records arenot transferred continuously as in database mirroring, but rathern intervals.Log shipping comprises backing up the transaction logs onhe primary server instance, copying the backed-up logs to thesecondary server instance, and restoring the copied logs on thesecondary server instance periodically. The primary databaseshould be configured for the Full Recovery or Bulk-Logged Recoverymodel. The secondary database should be initialized by a fullrecovery of the primary database with the NORECOVERY orSTANDBY option.Log shipping can optionally include a third server named “mon-itor” to record the history of log-shipping operations and to raisealerts during failures. Log-shipping operations are carried out byfour Microsoft SQL Server agent jobs: backup job, copy job, restorejob, and alert job.During log shipping, the secondary database is not completelysynchronized with the primary database. Also, log shipping doesnot provide automatic failover to the secondary server during afailure. Log shipping supports multiple secondary servers for aprimary server. In addition, a SQL Server instance can function asa secondary or monitor server for multiple primary servers. Thesecondary server can be used directly for limited query processing.Log shipping can be used in scenarios where multiple destina-tion servers are needed and where delays in restoring logs on theduplicate are acceptable.
Replication is a set of technologies for distributing data and data-base objects across Microsoft SQL Server databases over a net-work. Replication involves a publisher database instance thatmakes data available for copying, a distributor instance for copyingdata and maintaining metadata, and subscriber instances that canreceive replicated data.SQL Server 2005 provides three types of replication tech-nologies suitable for different needs: transactional, merge, andsnapshot. SQL Server 2005 adds enhancements to these repli-cation methods to help improve scalability, performance, andmonitoring capabilities. Some of the improvements includeadditional replication data types, support for partitionedtables and full-text index, direct Data Definition Language(DDL) replication for schema changes, and replication fromOracle
databases.In transactional replication, SQL Server 2005 adds an impor-tant feature called peer-to-peer replication, which enhances theexisting bidirectional replication option. Advantages of replicateddatabases include data load balancing and disaster recovery.With the peer-to-peer replication feature, administrators can setup multiple peer-to-peer transactional replication pairs amongdifferent data centers. That way, if one data center is down, itcan be failed over to its peer center(s). Although there is nolimit to the number of nodes in a peer-to-peer topology, man-ageability of the peer-to-peer relationships between nodes canbecome a constraint.
SQL Server 2005 providesthree types of replicationtechnologies suitable fordifferent needs: transactional,merge, and snapshot.SQL Server 2005 addsenhancements to thesereplication methods tohelp improve scalability,performance, and monitoringcapabilities.

Activity (3)

You've already reviewed this. Edit your review.
1 thousand reads
1 hundred reads

You're Reading a Free Preview

/*********** DO NOT ALTER ANYTHING BELOW THIS LINE ! ************/ var s_code=s.t();if(s_code)document.write(s_code)//-->