Displacing Backup is the New End Game for Replication Software
Replication software is increasingly entering the conversation as the logical replacement for backup software in client environments. Yet replicating data is really the easy part. Integrating the replication software so it becomes part of the fabric of a company's infrastructure is a far more difficult task. It is also one of the reasons that replication software has, to date, made so little headway in terms of displacing backup software for enterprise wide data protection. But as replication software matures, that will change.
I recently had a chance to speak with Peter O'Brien, InMage Systems Vice President of Business Development, about what features he now sees driving the adoption of InMage Systems' Scout in customer accounts. While Scout continues to have success as a business continuity solution, he sees more companies starting to use it to address other immediate business needs that they have. For instance, one company recently started to use Scout to measure and forecast data change rates in its environment.
But displacing backup software with replication software on servers is becoming the new end game for replication software. Replication software will never completely replace backup software as backup software is still needed to manage tape drives and tape libraries as well as the tape cartridges themselves. However replication software is certainly becoming a more viable option to displace backup software on the servers as more environments move towards disk as a backup target. Further, more organizations need near-zero backup windows and faster recovery times which are features that replication software, not backup software, is better equipped to provide.
So why have companies not adopted replication software already? The answer is simple according to O'Brien. Most replication software does not provide enough other functionality that organizations need to justify its wide-scale implementation. Features that he sees as replication software needing to do a better job of providing are the support of data movement, data migrations and greater application extensibility.
He notes that replication software received a big boost over the last few years with the introduction of continuous data protection (CDP) technology that tracks most or all changes to production data. But CDP alone has not been enough to push replication over the top. Backup software has decades of technology behind it, a management interface and policies that administrators are accustomed to using and deep integration with the many applications that it protects.
The other issue is that as organizational data stores continue to grow, administrators need simple and effective ways to migrate and move this data locally and remotely. Pure replication software may or may not handle this task very well or, if it does, it only moves or migrates for one server or one application, not every server in the enterprise. As a result, it is really not suitable for wide scale deployment.
But a number of things have changed in the last few years that now make replication software products like Scout viable for enterprise wide deployments including:
I recently had a chance to speak with Peter O'Brien, InMage Systems Vice President of Business Development, about what features he now sees driving the adoption of InMage Systems' Scout in customer accounts. While Scout continues to have success as a business continuity solution, he sees more companies starting to use it to address other immediate business needs that they have. For instance, one company recently started to use Scout to measure and forecast data change rates in its environment.
But displacing backup software with replication software on servers is becoming the new end game for replication software. Replication software will never completely replace backup software as backup software is still needed to manage tape drives and tape libraries as well as the tape cartridges themselves. However replication software is certainly becoming a more viable option to displace backup software on the servers as more environments move towards disk as a backup target. Further, more organizations need near-zero backup windows and faster recovery times which are features that replication software, not backup software, is better equipped to provide.
So why have companies not adopted replication software already? The answer is simple according to O'Brien. Most replication software does not provide enough other functionality that organizations need to justify its wide-scale implementation. Features that he sees as replication software needing to do a better job of providing are the support of data movement, data migrations and greater application extensibility.
He notes that replication software received a big boost over the last few years with the introduction of continuous data protection (CDP) technology that tracks most or all changes to production data. But CDP alone has not been enough to push replication over the top. Backup software has decades of technology behind it, a management interface and policies that administrators are accustomed to using and deep integration with the many applications that it protects.
The other issue is that as organizational data stores continue to grow, administrators need simple and effective ways to migrate and move this data locally and remotely. Pure replication software may or may not handle this task very well or, if it does, it only moves or migrates for one server or one application, not every server in the enterprise. As a result, it is really not suitable for wide scale deployment.
But a number of things have changed in the last few years that now make replication software products like Scout viable for enterprise wide deployments including:
- Improve data movement and migration capabilities. Scout can move and migrate data between different storage platforms so organizations can recover data locally or remotely.
- It has application level plug-ins. This means that as replication occurs, Scout prompts the applications to flush their buffers and create consistent recovery points. These consistent recovery points can be used for remote recoveries or even as a source for off-host backups.
- It can manage multiple replication jobs from a central console. Scout does not require administrators to log onto a specific console on each server in a replication pair. Rather it can manage replication and recovery for all servers performing replication from a central console which is needed for enterprise deployments.
I found this article interesting. In most DR scenarios for active/active or active/passive this is definitely the "best in class" direction for HA/RTOs...
There will always be certain DR scenarios that continue to require a means to restore "historical snapshots" of data. e.g., ETL process "corrupts" data structures due to errors in the source data system extracts that were unknown and now the data structures are unusable. Recovery from a "backup" of the data stores to fix is required and then once the ETL is corrected the "catchup" process to get the data currency commences...
make sense?
Cheers
I agree and disagree.
I think replication is the new way forward for backup. Traditional backup is horrible as it hammers the systems. Traditional backups offer very poor Recovery Point Objectives.
What I disagree with is the use of Volume Replication which I believe is what Inmage is doing.
Taking complete images of systems, have no indexing or granular retention is plain dumb. It also makes de-duplication harder (if they do it at all).
/tmp
In an era when vendors are doing things at the object level (ie files), volume replication is several step backwards.
Its great for DR but I cant imagine using this stuff to protect 1000 workstations. Can you? No, its not backup its the wrong type of technology.
Media Pundit,
Some of your assessments are correct and some are not.
InMage has the capability to do replication/CDP at the file level as they have an agent that operates at the file system level.
However, you do make a good point about data awareness and restoration of individual files.
In my conversations with them, I know it is something they are working on.
Jerome
Uhm... I dont think so. To quote from their own web site:
"By using a light-weight volume replication agent on the application server ..."
The agent also integrates with applications to synchronise and provide points of consistancy but I dont think it does much else.
Media Pundit,
InMage does actually have a file system agent though they do not always promote it or talk about it the way they should. InMage makes a passing reference to it on its web site:
http://www.inmage.net/practical-data-replication-for-todays-enterprise.html
and DCIG also talked about it about a year ago in the following blog entry:
http://inmage.dciginc.com/2008/05/inmage-protects-sharepoint-pt2.html
Jerome
Media Pundit,
Disclaimer: I still work at InMage.
DR-Scout is primarily intended for protecting mission critical applications, think ERP, Messaging, File Servers, Databases. It can also be used for protecting VDI type desktop consolidation environments. (vista/xp support)
If you only want to backup your vacation pictures on home desktops/laptops, then file based approaches are better suited. If you want to be able to restore your workstation to it's exact state after it died at work, then volume based approach is a better fit.
Just because you are protecting at a volume level in no way limits you from performing more granular recoveries, or for that matter granular retention. The product offers granular recoveries at fs and object (mail etc.) levels, and integrates well with both tape and disk based archival partner products at the smallest granularity suitable for compliance needs (files/emails).
Backing up at the volume level has the advantage of a small footprint, along with better fit with application consistency apis such as VSS which are volume based. Not to mention the possibility of doing off host backups when doing things at the volume level.