Isilon OneFS. Version 7.1. Backup and recovery guide

Transcription

1 Isilon OneFS Version 7.1 Backp and recovery gide

2 Copyright EMC Corporation. All rights reserved. Pblished in USA. Pblished March, 2014 EMC believes the information in this pblication is accrate as of its pblication date. The information is sbject to change withot notice. The information in this pblication is provided as is. EMC Corporation makes no representations or warranties of any kind with respect to the information in this pblication, and specifically disclaims implied warranties of merchantability or fitness for a particlar prpose. Use, copying, and distribtion of any EMC software described in this pblication reqires an applicable software license. EMC², EMC, and the EMC logo are registered trademarks or trademarks of EMC Corporation in the United States and other contries. All other trademarks sed herein are the property of their respective owners. For the most p-to-date reglatory docment for yor prodct line, go to EMC Online Spport ( For docmentation on EMC Data Domain prodcts, go to the EMC Data Domain Spport Portal ( EMC Corporation Hopkinton, Massachsetts In North America OneFS 7.1 Backp and recovery gide

3 CONTENTS Chapter 1 Introdction to this gide 7 Abot this gide...8 Isilon scale-ot NAS overview...8 Where to go for spport...8 Chapter 2 OneFS backp and recovery 9 OneFS backp and recovery overview...10 SyncIQ backp and recovery overview...10 NDMP backp and recovery overview Chapter 3 Data replication with SyncIQ 13 Replication policies and jobs Sorce and target clster association...15 Fll and differential replication...15 Controlling replication job resorce consmption Replication reports...16 Replication snapshots...16 Sorce clster snapshots...16 Target clster snapshots Data failover and failback with SyncIQ...17 Data failover Data failback...18 Replication and backp with SmartLock SmartLock replication and backp limitations...19 Recovery times and objectives for SyncIQ...20 SyncIQ license fnctionality Chapter 4 Backing p data with SyncIQ 21 Creating replication policies Exclding directories in replication...22 Exclding files in replication File criteria options Configre defalt replication policy settings...25 Create a replication policy...25 Create a SyncIQ domain Assess a replication policy Managing replication to remote clsters...31 Start a replication job...31 Pase a replication job...32 Resme a replication job...32 Cancel a replication job...32 View active replication jobs...32 View replication performance information Replication job information Managing failed replication jobs Resolve a replication policy...33 Reset a replication policy OneFS 7.1 Backp and recovery gide 3

4 CONTENTS Perform a fll or differential replication Managing replication policies Modify a replication policy Delete a replication policy...35 Enable or disable a replication policy View replication policies Replication policy information...36 Replication policy settings Managing replication to the local clster Cancel replication to the local clster...40 Break local target association View replication policies targeting the local clster Remote replication policy information Chapter 5 Recovering data with SyncIQ 43 Initiating data failover and failback with SyncIQ Fail over data to a secondary clster...44 Revert a failover operation Fail back data to a primary clster Performing disaster recovery for SmartLock directories...46 Recover SmartLock directories on a target clster...46 Migrate SmartLock directories...47 Chapter 6 NDMP backp 49 NDMP two way backp Snapshot-based incremental backps NDMP protocol spport Spported DMAs NDMP hardware spport NDMP backp limitations NDMP performance recommendations Exclding files and directories from NDMP backps Chapter 7 Backing p and recovering data with NDMP 57 NDMP backp and recovery tasks...58 Configring basic NDMP backp settings NDMP backp settings View NDMP backp settings...58 Configre and enable NDMP backp Disable NDMP backp...59 Managing NDMP ser acconts Create an NDMP ser accont View NDMP ser acconts...59 Modify the password of an NDMP ser accont Delete an NDMP ser accont Managing NDMP backp devices...60 NDMP backp device settings Detect NDMP backp devices...60 View NDMP backp devices Modify the name of an NDMP backp device...61 Delete an entry for an NDMP backp device Managing NDMP backp ports NDMP backp port settings OneFS 7.1 Backp and recovery gide

5 CONTENTS View NDMP backp ports Modify NDMP backp port settings Enable or disable an NDMP backp port...63 Managing NDMP backp sessions...63 NDMP session information View NDMP sessions...65 End an NDMP session Managing restartable backps Configre restartable backps View restartable backp contexts...66 Delete a restartable backp context Configre restartable backp settings View restartable backp settings Sharing tape drives between clsters Managing defalt NDMP settings...67 Set defalt NDMP settings for a directory Modify defalt NDMP settings for a directory View defalt NDMP settings for directories NDMP environment variables Managing snapshot based incremental backps...71 Enable snapshot-based incremental backps for a directory View snapshots for snapshot-based incremental backps Delete snapshots for snapshot-based incremental backps...71 View NDMP backp logs...72 Configring NDMP backps with EMC NetWorker...72 Create a grop Scan for tape devices Configre a library...73 Create a data media pool Label tape devices Create a metadata media pool Create a client Configring NDMP backp with Symantec NetBackp Add an NDMP host Configre storage devices Create a volme pool Inventory a robot...78 Create a NetBackp policy Configring NDMP backp with CommValt Simpana...80 Add a NAS client Add an NDMP library Create a storage policy...81 Assign a storage policy and schedle to a client...82 OneFS 7.1 Backp and recovery gide 5

6 CONTENTS 6 OneFS 7.1 Backp and recovery gide

7 CHAPTER 1 Introdction to this gide This section contains the following topics: Abot this gide...8 Isilon scale-ot NAS overview...8 Where to go for spport...8 Introdction to this gide 7

8 Introdction to this gide Abot this gide Isilon scale-ot NAS overview Where to go for spport This gide describes how to back p and recover data on Isilon clsters throgh either the SyncIQ software modle or the Network Data Management Protocol (NDMP). The EMC Isilon scale-ot NAS storage platform combines modlar hardware with nified software to harness nstrctred data. Powered by the distribted OneFS operating system, an EMC Isilon clster delivers a scalable pool of storage with a global namespace. The platform's nified software provides centralized web-based and command-line administration to manage the following featres: A symmetrical clster that rns a distribted file system Scale-ot nodes that add capacity and performance Storage options that manage files, block data, and tiering Flexible data protection and high availability Software modles that control costs and optimize resorces Yo can contact EMC Isilon Technical Spport for any qestions abot EMC Isilon prodcts. Online Spport Telephone Spport Help with online spport Live Chat Create a Service Reqest United States: (1-800-SVC-4EMC) Canada: Worldwide: For local phone nmbers in yor contry, see EMC Cstomer Spport Centers. For qestions specific to EMC Online Spport registration or access, spport@emc.com. 8 OneFS 7.1 Backp and recovery gide

9 CHAPTER 2 OneFS backp and recovery This section contains the following topics: OneFS backp and recovery overview...10 SyncIQ backp and recovery overview NDMP backp and recovery overview OneFS backp and recovery 9

10 OneFS backp and recovery OneFS backp and recovery overview Yo can back p data stored on Isilon clsters to another Isilon clster or a tape device. Yo can back p data to an Isilon clster with the SyncIQ software modle. SyncIQ enables yo to recover backed p data throgh failover and failback. Failover enables yo to access data on the clster it was backed p to. After yo fail over, yo can fail back to resme accessing yor data on the clster it was backed p from. Yo can back p data to a tape device over NDMP. After yo back p data to a tape device, yo can restore the data to any Isilon clster. SyncIQ backp and recovery overview OneFS enables yo to replicate data from one Isilon clster to another throgh the SyncIQ software modle. Yo mst activate a SyncIQ license on both Isilon clsters before yo can replicate data between them. Yo can replicate data at the directory level while optionally exclding specific files and sb-directories from being replicated. SyncIQ creates and references snapshots to replicate a consistent point-in-time image of a root directory. Metadata sch as access control lists (ACLs) and alternate data streams (ADS) are replicated along with data. SyncIQ enables yo to maintain a consistent backp copy of yor data on another Isilon clster. SyncIQ offers atomated failover and failback capabilities that enable yo to contine operations on another Isilon clster if a primary clster becomes navailable. NDMP backp and recovery overview In OneFS, yo can back p and restore file-system data throgh the Network Data Management Protocol (NDMP). From a backp server, yo can direct backp and recovery processes between an Isilon clster and backp devices sch as tape devices, media servers, and virtal tape libraries (VTLs). OneFS spports both NDMP three-way backp and NDMP two-way backp. Dring an NDMP three-way backp operation, a data management application (DMA) on a backp server instrcts the clster to start backing p data to a tape media server that is either attached to the LAN or directly attached to the DMA. Dring a two-way NDMP backp, a DMA on a backp server instrcts a Backp Accelerator node on the clster to start backing p data to a tape media server that is attached to the Backp Accelerator node. NDMP two-way backp is the most efficient method in terms of clster resorce consmption. However, NDMP two-way backp reqires that yo attach one or more Backp Accelerator nodes to the clster. In both the NDMP two-way and three-way backp models, file history data is transferred from the clster to the backp server. Before a backp begins, OneFS creates a snapshot of the targeted directory, then backs p the snapshot, which ensres that the backp image represents a specific point in time. Yo do not need to activate a SnapshotIQ license on the clster to perform NDMP backps. If yo have activated a SnapshotIQ license on the clster, yo can generate a snapshot throgh the SnapshotIQ tool, and then back p the same snapshot to mltiple tape devices. If yo back p a SnapshotIQ snapshot, OneFS does not create another snapshot for the backp. 10 OneFS 7.1 Backp and recovery gide

11 OneFS backp and recovery Note If yo are backing p SmartLock directories for compliance prposes, it is recommended that yo do not specify atocommit time periods for the SmartLock directories. This is becase, depending on the atocommit period, files in the SmartLock directories may still be sbject to change. NDMP backp and recovery overview 11

12

13 CHAPTER 3 Data replication with SyncIQ This section contains the following topics: Replication policies and jobs...14 Replication snapshots...16 Data failover and failback with SyncIQ...17 Replication and backp with SmartLock Recovery times and objectives for SyncIQ...20 SyncIQ license fnctionality Data replication with SyncIQ 13

14 Data replication with SyncIQ Replication policies and jobs Data replication is coordinated according to replication policies and jobs. Replication policies specify what data is replicated, where the data is replicated to, and how often the data is replicated. Replication jobs are the operations that replicate data from one Isilon clster to another. SyncIQ generates replication jobs according to replication policies. A replication policy specifies two clsters: the sorce and the target. The clster on which the replication policy exists is the sorce clster. The clster that data is being replicated to is the target clster. When a replication policy starts, SyncIQ generates a replication job for the policy. When a replication job rns, files from a directory on the sorce clster are replicated to a directory on the target clster; these directories are known as sorce and target directories. After the first replication job created by a replication policy finishes, the target directory and all files contained in the target directory are set to a read-only state, and can be modified only by other replication jobs belonging to the same replication policy. There is no limit to the nmber of replication policies that can exist on a clster. Note 14 OneFS 7.1 Backp and recovery gide To prevent permissions errors, make sre that ACL policy settings are the same across sorce and target clsters. Yo can create two types of replication policies: synchronization policies and copy policies. A synchronization policy maintains an exact replica of the sorce directory on the target clster. If a file or sb-directory is deleted from the sorce directory, the file or directory is deleted from the target clster when the policy is rn again. Yo can se synchronization policies to fail over and fail back data between sorce and target clsters. When a sorce clster becomes navailable, yo can fail over data on a target clster and make the data available to clients. When the sorce clster becomes available again, yo can fail back the data to the sorce clster. A copy policy maintains recent versions of the files that are stored on the sorce clster. However, files that are deleted on the sorce clster are not deleted from the target clster. Failback is not spported for copy policies. Copy policies are most commonly sed for archival prposes. Copy policies enable yo to remove files from the sorce clster withot losing those files on the target clster. Deleting files on the sorce clster improves performance on the sorce clster while maintaining the deleted files on the target clster. This can be sefl if, for example, yor sorce clster is being sed for prodction prposes and yor target clster is being sed only for archiving. After creating a job for a replication policy, SyncIQ mst wait ntil the job completes before it can create another job for the policy. Any nmber of replication jobs can exist on a clster at a given time; however, only five replication jobs can rn on a sorce clster at the same time. If more than five replication jobs exist on a clster, the first five jobs rn while the others are qeed to rn. The nmber of replication jobs that a single target clster can spport concrrently is dependent on the nmber of workers available on the target clster. Yo can replicate any nmber of files and directories with a single replication job. Yo can prevent a large replication job from overwhelming the system by limiting the amont of clster resorces and network bandwidth that data synchronization is allowed to consme. Becase each node in a clster is able to send and receive data, the speed at which data is replicated increases for larger clsters.

15 Data replication with SyncIQ Sorce and target clster association SyncIQ associates a replication policy with a target clster by marking the target clster when the job rns for the first time. Even if yo modify the name or IP address of the target clster, the mark persists on the target clster. When a replication policy is rn, SyncIQ checks the mark to ensre that data is being replicated to the correct location. On the target clster, yo can manally break an association between a replication policy and target directory. Breaking the association between a sorce and target clster cases the mark on the target clster to be deleted. Yo might want to manally break a target association if an association is obsolete. If yo break the association of a policy, the policy is disabled on the sorce clster and yo cannot rn the policy. If yo want to rn the disabled policy again, yo mst reset the replication policy. Note Fll and differential replication Breaking a policy association cases either a fll or differential replication to occr the next time yo rn the replication policy. Dring a fll or differential replication, SyncIQ creates a new association between the sorce and target clsters. Depending on the amont of data being replicated, a fll or differential replication can take a very long time to complete. If a replication policy enconters an isse that cannot be fixed (for example, if the association was broken on the target clster), yo might need to reset the replication policy. If yo reset a replication policy, SyncIQ performs either a fll or differential replication the next time the policy is rn. Yo can specify the type of replication that SyncIQ performs. Dring a fll replication, SyncIQ transfers all data from the sorce clster regardless of what data exists on the target clster. A fll replication consmes large amonts of network bandwidth and can take a very long time to complete. However, a fll replication is less strenos on CPU sage than a differential replication. Dring a differential replication, SyncIQ first checks whether a file already exists on the target clster and then transfers only data that does not already exist on the target clster. A differential replication consmes less network bandwidth than a fll replication; however, differential replications consme more CPU. Differential replication can be mch faster than a fll replication if there is an adeqate amont of available CPU for the differential replication job to consme. Controlling replication job resorce consmption Yo can create rles that limit the network traffic created and the rate at which files are sent by replication jobs. Yo can also specify the nmber of workers that are spawned by a replication policy to limit the amont of clster resorces that are consmed. Also, yo can restrict a replication policy to connect only to a specific storage pool. Yo can create network-traffic rles that control the amont of network traffic generated by replication jobs dring specified time periods. These rles can be sefl if, for example, yo want to limit the amont of network traffic created dring other resorceintensive operations. Yo can create mltiple network traffic rles to enforce different limitations at different times. For example, yo might allocate a small amont of network bandwidth dring peak bsiness hors, bt allow nlimited network bandwidth dring non-peak hors. Sorce and target clster association 15

16 Data replication with SyncIQ When a replication job rns, OneFS generates workers on the sorce and target clster. Workers on the sorce clster send data while workers on the target clster write data. OneFS generates no more than 40 workers for a replication job. Yo can modify the maximm nmber of workers generated per node to control the amont of resorces that a replication job is allowed to consme. For example, yo can increase the maximm nmber of workers per node to increase the speed at which data is replicated to the target clster. Yo can also redce resorce consmption throgh file-operation rles that limit the rate at which replication policies are allowed to send files. However, it is recommended that yo only create file-operation rles if the files yo intend to replicate are predictably similar in size and not especially large. Replication reports After a replication job completes, SyncIQ generates a report that contains detailed information abot the job, inclding how long the job ran, how mch data was transferred, and what errors occrred. If a replication report is interrpted, SyncIQ might create a sbreport abot the progress of the job so far. If the job is then restarted, SyncIQ creates another sbreport abot the progress of the job ntil the job either completes or is interrpted again. SyncIQ creates a sbreport each time the job is interrpted ntil the job completes sccessflly. If mltiple sbreports are created for a job, SyncIQ combines the information from the sbreports into a single report. SyncIQ rotinely deletes replication reports. Yo can specify the maximm nmber of replication reports that SyncIQ retains and the length of time that SyncIQ retains replication reports. If the maximm nmber of replication reports is exceeded on a clster, SyncIQ deletes the oldest report each time a new report is created. Yo cannot cstomize the content of a replication report. Note If yo delete a replication policy, SyncIQ atomatically deletes any reports that were generated for that policy. Replication snapshots Sorce clster snapshots SyncIQ generates snapshots to facilitate replication, failover, and failback between Isilon clsters. Snapshots generated by SyncIQ can also be sed for archival prposes on the target clster. SyncIQ generates snapshots on the sorce clster to ensre that a consistent point-intime image is replicated and that naltered data is not sent to the target clster. Before rnning a replication job, SyncIQ creates a snapshot of the sorce directory. SyncIQ then replicates data according to the snapshot rather than the crrent state of the clster, allowing sers to modify sorce-directory files while ensring that an exact pointin-time image of the sorce directory is replicated. For example, if a replication job of /ifs/data/dir/ starts at 1:00 PM and finishes at 1:20 PM, and /ifs/data/dir/file is modified at 1:10 PM, the modifications are not reflected on the target clster, even if /ifs/data/dir/file is not replicated ntil 1:15 PM. 16 OneFS 7.1 Backp and recovery gide

17 Data replication with SyncIQ Target clster snapshots Yo can replicate data according to a snapshot generated with the SnapshotIQ tool. If yo replicate data according to a SnapshotIQ snapshot, SyncIQ does not generate another snapshot of the sorce directory. This method can be sefl if yo want to replicate identical copies of data to mltiple Isilon clsters. SyncIQ generates sorce snapshots to ensre that replication jobs do not transfer nmodified data. When a job is created for a replication policy, SyncIQ checks whether it is the first job created for the policy. If it is not the first job created for the policy, SyncIQ compares the snapshot generated for the earlier job with the snapshot generated for the new job. SyncIQ replicates only data that has changed since the last time a snapshot was generated for the replication policy. When a replication job is completed, SyncIQ deletes the previos sorce-clster snapshot and retains the most recent snapshot ntil the next job is rn. When a replication job is rn, SyncIQ generates a snapshot on the target clster to facilitate failover operations. When the next replication job is created for the replication policy, the job creates a new snapshot and deletes the old one. If a SnapshotIQ license has been activated on the target clster, yo can configre a replication policy to generate additional snapshots that remain on the target clster even as sbseqent replication jobs rn. SyncIQ generates target snapshots to facilitate failover on the target clster regardless of whether a SnapshotIQ license has been configred on the target clster. Failover snapshots are generated when a replication job completes. SyncIQ retains only one failover snapshot per replication policy, and deletes the old snapshot after the new snapshot is created. If a SnapshotIQ license has been activated on the target clster, yo can configre SyncIQ to generate archival snapshots on the target clster that are not atomatically deleted when sbseqent replication jobs rn. Archival snapshots contain the same data as the snapshots that are generated for failover prposes. However, yo can configre how long archival snapshots are retained on the target clster. Yo can access archival snapshots the same way that yo access other snapshots generated on a clster. Data failover and failback with SyncIQ SyncIQ enables yo to perform atomated data failover and failback operations between Isilon clsters. If a clster is rendered nsable, yo can fail over to another Isilon clster, enabling clients to access to access their data on the other clster. If the nsable clster becomes accessible again, yo can fail back to the original Isilon clster. For the prposes of explaining failover and failback procedres, the clster originally accessed by clients is referred to as the primary clster, and the clster that client data is originally replicated to is referred to as the secondary clster. Failover is the process that allows clients to modify data on a secondary clster. Failback is the process that allows clients to access data on the primary clster again and begins to replicate data back to the secondary clster. Failover and failback can be sefl in disaster recovery procedres. For example, if a primary clster is damaged by a natral disaster, yo can migrate clients to a secondary clster ntil the primary clster is repaired and then migrate the clients back to the primary clster. Target clster snapshots 17

18 Data replication with SyncIQ Yo can fail over and fail back to facilitate schedled clster maintenance. For example, if yo are pgrading the primary clster, yo might want to migrate clients to a secondary clster ntil the pgrade is complete and then migrate clients back to the primary clster. Note Data failover and failback is not spported for SmartLock directories. Data failover Data failback Data failover is the process of preparing data on a secondary clster to be modified by clients. After yo fail over to a secondary clster, yo can redirect clients to modify their data on the secondary clster. Before failover is performed, yo mst create and rn a replication policy on the primary clster. Yo initiate the failover process on the secondary clster. Failover is performed per replication policy; to migrate data that is spread across mltiple replication policies, yo mst initiate failover for each replication policy. Yo can se any replication policy to fail over. However, if the action of the replication policy is set to copy, any file that was deleted on the primary clster will be present on the secondary clster. When the client connects to the secondary clster, all files that were deleted on the primary clster will be available to the client. If yo initiate failover for a replication policy while an associated replication job is rnning, the failover operation completes bt the replication job fails. Becase data might be in an inconsistent state, SyncIQ ses the snapshot generated by the last sccessfl replication job to revert data on the secondary clster to the last recovery point. If a disaster occrs on the primary clster, any modifications to data that were made after the last sccessfl replication job started are not reflected on the secondary clster. When a client connects to the secondary clster, their data appears as it was when the last sccessfl replication job was started. Data failback is the process of restoring clsters to the roles they occpied before a failover operation. After data failback is complete, the primary clster hosts clients and replicates data to the secondary clster for backp. The first step in the failback process is pdating the primary clster with all of the modifications that were made to the data on the secondary clster. The next step in the failback process is preparing the primary clster to be accessed by clients. The final step in the failback process is resming data replication from the primary to the secondary clster. At the end of the failback process, yo can redirect sers to resme accessing their data on the primary clster. Yo can fail back data with any replication policy that meets all of the following criteria: The sorce directory is not a SmartLock directory. The policy has been failed over. The policy is a synchronization policy. The policy does not exclde any files or directories from replication. 18 OneFS 7.1 Backp and recovery gide

19 Data replication with SyncIQ Replication and backp with SmartLock Yo mst ensre that SmartLock directories remain protected dring replication and backp operations. If yo are replicating SmartLock directories with SyncIQ, it is recommended that yo configre all nodes on the sorce and target clsters into Network Time Protocol (NTP) peer mode to ensre that the node clocks are synchronized. For compliance clsters, it is recommended that yo configre all nodes on the sorce and target clsters into NTP peer mode before yo set the compliance clock to ensre that the compliance clocks are initially set to the same time. Note Do not configre SmartLock settings for a target SmartLock directory nless yo are no longer replicating data to the directory. Configring an atocommit time period for a target SmartLock directory can case replication jobs to fail. If the target SmartLock directory commits a file to a WORM state, and the file is modified on the sorce clster, the next replication job will fail becase it cannot pdate the file. SmartLock replication and backp limitations Be aware of the limitations of replicating and backing p SmartLock directories with SyncIQ and NDMP. If the sorce or target directory of a SyncIQ policy is a SmartLock directory, replication might not be allowed. For more information, see the following table: Sorce directory type Target directory type Allowed Non-SmartLock Non-SmartLock Yes Non-SmartLock SmartLock enterprise Yes Non-SmartLock SmartLock compliance No SmartLock enterprise Non-SmartLock Yes; however, retention dates and commit stats of files will be lost. SmartLock enterprise SmartLock enterprise Yes SmartLock enterprise SmartLock compliance No SmartLock compliance Non-SmartLock No SmartLock compliance SmartLock enterprise No SmartLock compliance SmartLock compliance Yes If yo replicate SmartLock directories to another clster with SyncIQ, the WORM state of files is replicated. However, SmartLock directory configration settings are not transferred to the target directory. For example, if yo replicate a directory that contains a committed file that is set to expire on March 4th, the file is still set to expire on March 4th on the target clster. However, if the directory on the sorce clster is set to prevent files from being committed for more than a year, the target directory is not atomatically set to the same restriction. Replication and backp with SmartLock 19

20 Data replication with SyncIQ If yo back p data to an NDMP device, all SmartLock metadata relating to the retention date and commit stats is transferred to the NDMP device. If yo restore data to a SmartLock directory on the clster, the metadata persists on the clster. However, if the directory that yo restore to is not a SmartLock directory, the metadata is lost. Yo can restore to a SmartLock directory only if the directory is empty. Recovery times and objectives for SyncIQ SyncIQ license fnctionality The Recovery Point Objective (RPO) and the Recovery Time Objective (RTO) are measrements of the impacts that a disaster can have on bsiness operations. Yo can calclate yor RPO and RTO for a disaster recovery with replication policies. RPO is the maximm amont of time for which data is lost if a clster sddenly becomes navailable. For an Isilon clster, the RPO is the amont of time that has passed since the last completed replication job started. The RPO is never greater than the time it takes for two consective replication jobs to rn and complete. If a disaster occrs while a replication job is rnning, the data on the secondary clster is reverted to the state it was in when the last replication job completed. For example, consider an environment in which a replication policy is schedled to rn every three hors, and replication jobs take two hors to complete. If a disaster occrs an hor after a replication job begins, the RPO is for hors, becase it has been for hors since a completed job began replicating data. RTO is the maximm amont of time reqired to make backp data available to clients after a disaster. The RTO is always less than or approximately eqal to the RPO, depending on the rate at which replication jobs are created for a given policy. If replication jobs rn continosly, meaning that another replication job is created for the policy before the previos replication job completes, the RTO is approximately eqal to the RPO. When the secondary clster is failed over, the data on the clster is reset to the state it was in when the last job completed; resetting the data takes an amont of time proportional to the time it took sers to modify the data. If replication jobs rn on an interval, meaning that there is a period of time after a replication job completes before the next replication job for the policy starts, the relationship between RTO and RPO depends on whether a replication job is rnning when the disaster occrs. If a job is in progress when a disaster occrs, the RTO is roghly eqal to the RPO. However, if a job is not rnning when a disaster occrs, the RTO is negligible becase the secondary clster was not modified since the last replication job ran, and the failover process is almost instantaneos. Yo can replicate data to another Isilon clster only if yo activate a SyncIQ license on both the local clster and the target clster. If a SyncIQ license becomes inactive, yo cannot create, rn, or manage replication policies. Also, all previosly created replication policies are disabled. Replication policies that target the local clster are also disabled. However, data that was previosly replicated to the local clster is still available. 20 OneFS 7.1 Backp and recovery gide

21 CHAPTER 4 Backing p data with SyncIQ This section contains the following topics: Creating replication policies Managing replication to remote clsters...31 Managing failed replication jobs Managing replication policies...35 Managing replication to the local clster Backing p data with SyncIQ 21

22 Backing p data with SyncIQ Creating replication policies Exclding directories in replication Yo can create replication policies that determine when data is replicated with SyncIQ. Yo can exclde directories from being replicated by replication policies even if the directories exist nder the specified sorce directory. Note Yo cannot fail back replication policies that exclde directories. By defalt, all files and directories nder the sorce directory of a replication policy are replicated to the target clster. However, yo can prevent directories nder the sorce directory from being replicated. If yo specify a directory to exclde, files and directories nder the exclded directory are not replicated to the target clster. If yo specify a directory to inclde, only the files and directories nder the inclded directory are replicated to the target clster; any directories that are not contained in an inclded directory are exclded. If yo both inclde and exclde directories, any exclded directories mst be contained in one of the inclded directories; otherwise, the exclded-directory setting has no effect. For example, consider a policy with the following settings: The root directory is /ifs/data The inclded directories are /ifs/data/media/msic and /ifs/data/ media/movies The exclded directories are /ifs/data/archive and /ifs/data/media/ msic/working In this example, the setting that excldes the /ifs/data/archive directory has no effect becase the /ifs/data/archive directory is not nder either of the inclded directories. The /ifs/data/archive directory is not replicated regardless of whether the directory is explicitly exclded. However, the setting that excldes the /ifs/data/ media/msic/working directory does have an effect, becase the directory wold be replicated if the setting was not specified. In addition, if yo exclde a directory that contains the sorce directory, the excldedirectory setting has no effect. For example, if the root directory of a policy is /ifs/ data, explicitly exclding the /ifs directory does not prevent /ifs/data from being replicated. Any directories that yo explicitly inclde or exclde mst be contained in or nder the specified root directory. For example, consider a policy in which the specified root directory is /ifs/data. In this example, yo cold inclde both the /ifs/data/ media and the /ifs/data/sers/ directories becase they are nder /ifs/data. Exclding directories from a synchronization policy does not case the directories to be deleted on the target clster. For example, consider a replication policy that synchronizes /ifs/data on the sorce clster to /ifs/data on the target clster. If the policy excldes /ifs/data/media from replication, and /ifs/data/media/ file exists on the target clster, rnning the policy does not case /ifs/data/ media/file to be deleted from the target clster. 22 OneFS 7.1 Backp and recovery gide

23 Backing p data with SyncIQ Exclding files in replication If yo do not want specific files to be replicated by a replication policy, yo can exclde them from the replication process throgh file-matching criteria statements. Yo can configre file-matching criteria statements dring the replication-policy creation process. Note Yo cannot fail back replication policies that exclde files. A file-criteria statement can inclde one or more elements. Each file-criteria element contains a file attribte, a comparison operator, and a comparison vale. Yo can combine mltiple criteria elements in a criteria statement with Boolean "AND" and "OR" operators. Yo can configre any nmber of file-criteria definitions. Configring file-criteria statements can case the associated jobs to rn slowly. It is recommended that yo specify file-criteria statements in a replication policy only if necessary. Modifying a file-criteria statement will case a fll replication to occr the next time that a replication policy is started. Depending on the amont of data being replicated, a fll replication can take a very long time to complete. For synchronization policies, if yo modify the comparison operators or comparison vales of a file attribte, and a file no longer matches the specified file-matching criteria, the file is deleted from the target the next time the job is rn. This rle does not apply to copy policies. File criteria options Yo can configre a replication policy to exclde files that meet or do not meet specific criteria. Yo can specify file criteria based on the following file attribtes: Date created Incldes or excldes files based on when the file was created. This option is available for copy policies only. Yo can specify a relative date and time, sch as "two weeks ago", or specific date and time, sch as "Janary 1, 2012." Time settings are based on a 24-hor clock. Date accessed Incldes or excldes files based on when the file was last accessed. This option is available for copy policies only, and only if the global access-time-tracking option of the clster is enabled. Yo can specify a relative date and time, sch as "two weeks ago", or specific date and time, sch as "Janary 1, 2012." Time settings are based on a 24-hor clock. Date modified Incldes or excldes files based on when the file was last modified. This option is available for copy policies only. Yo can specify a relative date and time, sch as "two weeks ago", or specific date and time, sch as "Janary 1, 2012." Time settings are based on a 24-hor clock. Exclding files in replication 23

24 Backing p data with SyncIQ File name Incldes or excldes files based on the file name. Yo can specify to inclde or exclde fll or partial names that contain specific text. The following wildcard characters are accepted: Note Alternatively, yo can filter file names by sing POSIX reglar-expression (regex) text. Isilon clsters spport IEEE Std (POSIX.2) reglar expressions. For more information abot POSIX reglar expressions, see the BSD man pages. Table 1 Replication file matching wildcards Wildcard Description - - * Matches any string in place of the asterisk. For example, m* matches movies and m123. [ ] Matches any characters contained in the brackets, or a range of characters separated by a dash. For example, b[aei]t matches bat, bet, and bit. For example, 1[4-7]2 matches 142, 152, 162, and 172. Yo can exclde characters within brackets by following the first bracket with an exclamation mark. For example, b[!ie] matches bat bt not bit or bet. Yo can match a bracket within a bracket if it is either the first or last character. For example, [[c]at matches cat and [at. Yo can match a dash within a bracket if it is either the first or last character. For example, car[-s] matches cars and car-.? Matches any character in place of the qestion mark. For example, t?p matches tap, tip, and top. Path Incldes or excldes files based on the file path. This option is available for copy policies only. Yo can specify to inclde or exclde fll or partial paths that contain specified text. Yo can also inclde the wildcard characters *,?, and [ ]. Size Incldes or excldes files based on their size. Note File sizes are represented in mltiples of 1024, not OneFS 7.1 Backp and recovery gide

25 Backing p data with SyncIQ Type Incldes or excldes files based on one of the following file-system object types: Soft link Reglar file Directory Configre defalt replication policy settings Yo can configre defalt settings for replication policies. If yo do not modify these settings when creating a replication policy, the specified defalt settings are applied. Procedre 1. Click Data Protection > SyncIQ > Settings. 2. In the Defalt Policy Settings section, specify how yo want replication policies to connect to target clsters by selecting one of the following options: Click Connect to any nodes in the clster. Click Connect to only the nodes in the sbnet and pool if the target clster name specifies a SmartConnect zone. 3. Specify which nodes yo want replication policies to connect to when a policy is rn. Options Connect policies to all nodes on a sorce clster. Connect policies only to nodes contained in a specified sbnet and pool. Description Click Rn the policy on all nodes in this clster. a. Click Rn the policy only on nodes in the specified sbnet and pool. b. From the Sbnet and pool list, select the sbnet and pool. Note Create a replication policy SyncIQ does not spport dynamically allocated IP address pools. If a replication job connects to a dynamically allocated IP address, SmartConnect might reassign the address while a replication job is rnning, which wold disconnect the job and case it to fail. 4. Click Sbmit. Yo can create a replication policy with SyncIQ that defines how and when data is replicated to another Isilon clster. Configring a replication policy is a five-step process. Configre replication policies careflly. If yo modify any of the following policy settings after the policy is rn, OneFS performs either a fll or differential replication the next time the policy is rn: Sorce directory Inclded or exclded directories File-criteria statement Configre defalt replication policy settings 25

26 Backing p data with SyncIQ Target clster name or address This applies only if yo target a different clster. If yo modify the IP or domain name of a target clster, and then modify the replication policy on the sorce clster to match the new IP or domain name, a fll replication is not performed. Target directory Configre basic policy settings Yo mst configre basic settings for a replication policy. Procedre 1. Click Data Protection > SyncIQ > Policies. 2. Click Create a SyncIQ policy. 3. In the Settings area, in the Policy name field, type a name for the replication policy. 4. Optional: In the Description field, type a description for the replication policy. 5. In the Action area, specify the type of replication policy. To copy all files from the sorce directory to the target directory, click Copy. Note Failback is not spported for copy policies. To copy all files from the sorce directory to the target directory and delete any files on the target directory that are not in the sorce directory, click Synchronize. 6. In the Rn job area, specify whether replication jobs will be rn. Options Rn jobs only when manally initiated by a ser. Rn jobs atomatically according to a schedle. Rn jobs atomatically every time a change is made to the sorce directory. Description Click Only manally. a. Click On a schedle. b. Specify a schedle. If yo configre a replication policy to rn more than once a day, yo cannot configre the interval to span across two calendar days. For example, yo cannot configre a replication policy to rn every hor starting at 7:00 PM and ending at 1:00 AM. Click Whenever the sorce is modified. After yo finish The next step in the process of creating a replication policy is specifying sorce directories and files. 26 OneFS 7.1 Backp and recovery gide

27 Backing p data with SyncIQ Specify sorce directories and files Yo mst specify the directories and files yo want to replicate. Procedre 1. In the Sorce Clster area, in the Sorce Root Directory field, type the fll path of the sorce directory that yo want to replicate to the target clster. Yo mst specify a directory contained in /ifs. Yo cannot specify the / ifs/.snapshot directory or sbdirectory of it. 2. Optional: Prevent specific sbdirectories of the root directory from being replicated. To inclde a directory, in the Inclded Directories area, click Add a directory path. To exclde a directory, in the Exclded Directories area, click Add a directory path. 3. Optional: Prevent specific files from being replicated by specifying file matching criteria. a. In the File Matching Criteria area, select a filter type. b. Select an operator. c. Type a vale. Files that do not meet the specified criteria will not be replicated to the target clster. For example, if yo specify File Type doesn't match.txt, SyncIQ will not replicate any files with the.txt file extension. If yo specify Created after 08/14/2013, SyncIQ will not replicate any files created before Agst 14th, If yo want to specify more than one file matching criterion, yo can control how the criteria relate to each other by clicking either Add an "Or" condition or Add an "And" condition. 4. Specify which nodes yo want the replication policy to connect to when the policy is rn. Options Connect the policy to all nodes in the sorce clster. Connect the policy only to nodes contained in a specified sbnet and pool. Description Click Rn the policy on all nodes in this clster. a. Click Rn the policy only on nodes in the specified sbnet and pool. b. From the Sbnet and pool list, select the sbnet and pool. Note SyncIQ does not spport dynamically allocated IP address pools. If a replication job connects to a dynamically allocated IP address, SmartConnect might reassign the address while a replication job is rnning, which wold disconnect the job and case it to fail. After yo finish The next step in the process of creating a replication policy is specifying the target directory. Create a replication policy 27

28 Backing p data with SyncIQ Specify the policy target directory Yo mst specify a target clster and directory to replicate data to. Procedre 1. In the Target Clster area, in the Target Host field, type one of the following: The flly qalified domain name of any node in the target clster. The host name of any node in the target clster. The name of a SmartConnect zone in the target clster. The IPv4 or IPv6 address of any node in the target clster. localhost Note This will replicate data to another directory on the local clster. SyncIQ does not spport dynamically allocated IP address pools. If a replication job connects to a dynamically allocated IP address, SmartConnect might reassign the address while a replication job is rnning, which wold disconnect the job and case it to fail. 2. In the Target Directory field, type the absolte path of the directory on the target clster that yo want to replicate data to. CAUTION If yo specify an existing directory on the target clster, ensre that the directory is not the target of another replication policy. If this is a synchronization policy, ensre that the directory is empty. All files are deleted from the target of a synchronization policy the first time the policy is rn. If the specified target directory does not already exist on the target clster, the directory is created the first time the job is rn. It is recommended that yo do not specify the /ifs directory. If yo specify the /ifs directory, the entire target clster is set to a read-only state, preventing yo from storing any other data on the clster. If this is a copy policy, and files in the target directory share the same name as files in the sorce directory, the target directory files are overwritten when the job is rn. 3. If yo want replication jobs to connect only to the nodes inclded in the SmartConnect zone specified by the target clster, click Connect only to the nodes within the target clster SmartConnect Zone. After yo finish Configre policy target snapshot settings The next step in the process of creating a replication policy is specifying policy target snapshot settings. Yo can optionally specify how archival snapshots are generated on the target clster. Yo can access archival snapshots the same way that yo access SnapshotIQ snapshots. SyncIQ always retains one snapshot on the target clster to facilitate failover, regardless of these settings. 28 OneFS 7.1 Backp and recovery gide

29 Backing p data with SyncIQ Procedre 1. To create archival snapshots on the target clster, in the Target Snapshots area, click Captre snapshots on the target clster. 2. Optional: To modify the defalt alias of the last snapshot created according to the replication policy, in the Snapshot Alias Name field, type a new alias. Yo can specify the alias name as a snapshot naming pattern. For example, the following naming pattern is valid: %{PolicyName}-on-%{SrcClster}-latest The previos example prodces names similar to the following: newpolicy-on-clster1-latest 3. Optional: To modify the snapshot naming pattern for snapshots created according to the replication policy, in the Snapshot Naming Pattern field, type a naming pattern. Each snapshot generated for this replication policy is assigned a name based on this pattern. For example, the following naming pattern is valid: %{PolicyName}-from-%{SrcClster}-at-%H:%M-on-%m-%d-%Y The example prodces names similar to the following: newpolicy-from-clster1-at-10:30-on Select one of the following options: Click Snapshots do not expire. Click Snapshots expire after... and specify an expiration period. After yo finish Configre advanced policy settings The next step in the process of creating a replication policy is configring advanced policy settings. Yo can optionally configre advanced settings for a replication policy. Procedre 1. Optional: In the Worker Threads Per Node field, specify the maximm nmber of concrrent processes per node that will perform replication operations. Note Do not modify the defalt setting withot conslting Isilon Technical Spport. 2. Optional: From the Log Level list, select the level of logging yo want SyncIQ to perform for replication jobs. The following log levels are valid, listed from least to most verbose: Click Error. Click Notice. Click Network Activity. Click File Activity. 3. Optional: If yo want SyncIQ to perform a checksm on each file data packet that is affected by the replication policy, select the Validate File Integrity check box. If yo enable this option, and the checksm vales for a file data packet do not match, SyncIQ retransmits the affected packet. Create a replication policy 29