Asynchronous Replication with NexentaStor

This document discusses variousasynchronous replication options with NexentaStor. NexentaStor provides twodifferent replication services, AutoTier and AutoSync.

Both AutoSync and AutoTier supportincremental data transfers. After the initial replication, only changed data onthe source is sent for subsequent transfers.

AutoTier

AutoTier typically uses rsync as the data transferprotocol. Typical rsync options used by AutoTier are “-HlptgoD”.

AutoSync

The AutoSync service requires that bothsource and target are ZFS-based systems. Depending on the options used, theremay be further restrictions. For example, the “RR” protocol is supported onlybetween two NexentaStor systems.

Scheduler

NexentaStor includes a sophisticatedscheduler with its replication service. Replication can be scheduled for aspecific time, and run once or periodically. Some examples are:

o  Run at 3am

o  Run every 4 hours

o  Run once a week

o  Run right now

o  Run once, at 5pm.

The smallest time interval you canconfigure in NMV is 1 minute. With NMC you can configure AutoSync to run indaemon mode and do transfers every few seconds.

Controlling which files arereplicated

NexentaStor provides the ability to specifyfiles to not be included in the replication. This can be indicated whencreating or editing the replication service. To do this select More Options anduse the Exclude folders option. Here you specify patterns of files that you donot want to be included in the data transfer. For example, here is how youwould specify to not send files ending in “.jsp”.

For multiple patterns, please type as    *.jsp\|*.class\|*.properties

Specifying the Target System

NexentaStor can provide replicationservices to and from a variety of storage systems. However not all

Bi-directional Replication

NexentaStor can allow for replication to beperformed bi-directionally. A separate replication serviceis configured for each direction.

For bi-directional replication you need touse the AutoTier service.

Various usage models for bi-directionalreplication might be:

o  Add/update files only if theyare newer than the receiver

o  Add/update files only if theydon’t already exist on the receiver

o  Add/update files only if theyexist on the receiver

Add/update files only if they are newer than receiver

For this use case both the source andtarget storage systems are updating files in the same file systems. To resolveconflicts if the file is edited in both places, the file changed most recentlyis kept. Thus we will rely on the file modification times and it is criticalthat the source and target systems be using a time service such as NTP.

We need to use AutoTier for this use case because AutoSync makes certainassumptions about the target directories not being in use or modified.But AutoTier by default assumes that the source is the master and thereis no other unique activity happening on the target. So for this use casewe need to make some changes to the default rsync command-line options. To editthe rsync parameters, you need to select More Options when configuring theservice. This will cause the Window of available options to expand, as shown inthe following screenshot.

We need to make the following changes tothe default rsync options:

Remove:

--delete: If we use this option, then anyadditional files created by the receiver will be deleted.

--inplace: this doesn’t seem to be a goodoption if the receiver might be editing the files

Add:

--update: this will skip replication forfiles that are newer on the receiver

--times: preserve modification times. Thisoption may not be necessary as they seem to already be preserved.

When configuring the service, choose“Replicate content”. Select the source folder.

For the folder on the target system,specify the wildcard at the end of the target directory. For example, if thetarget is a remote system then you might reference the folder like this:“/mytarget/*”. If you don’t use the “Replicate content”option you might end upwith additional directories getting created.

For receiving data from the target system,you need to again specify the wildcard for the receiving directory. Howeverthis can’t be done using NMV. Instead, create the service first in NMC using“setup auto-tier”. This will allow you to specify the wildcard for the targetfolder on the local system. After the service is created in NMC, you can thenedit the rsync options using NMV as shown above.

Monitoring

The replication service is an importantmechanism for data protection. It is important to be able to check anddetermine if the service is functioning correctly. This can be done easily fromNMV. On the summary view for the services, there is an option to View Log:

After selecting View log, you may be askedto provided the administrative password if you are not already logged in. Nexta dialog box appears showing you the log of past executions of the replicationservice and whether the task was successful.

Note that by default NexentaStor willre-run the service automatically if the service failed due to a loss ofconnectivity. This can be changed by selecting “More Options” when creating orediting the service.

Related Topics

AutoCDP provides synchronous replication.The user manual can be found at http://www.nexenta.com/corp/documentation/product-documentation.

The rsync protocol options are describedhere: http://rsync.samba.org/ftp/rsync/rsync.html.