Tony's blog

Voice of the Customer: Collection Replication, Disaster Recovery and Tape Elimination

The following is a discussion I had with a Data Domain customer - Rob Carlisi, Systems Engineer for Orange County Public Schools (OCPS). 

 

You can also listen to the podcast - Download Interview Orange County Public Schools

 

A couple of things to point out:

• OCPS has eliminated using tapes completely. 

• They are using Data Domain Collection Replication for their off-site disaster recovery.

Tony Asaro: Rob, welcome and I appreciate you participating this discussion.  Please give us a background on your company.

 

Rob Carlisi: I work at Orange County Public Schools in Orlando, Florida. We’re the 11^thlargest school district in the country. The amount of online storage that we have is a little over 100 terabytes and the amount of backup data we have on Data Domain is about 450 terabytes.

 

Tony Asaro: And what is your role with the company, Rob?

 

Rob Carlisi: I’m a systems engineer and I’ve been here for a little over 6 years. Our group is primarily responsible for anything that happens inside the data center that’s not network related. I’m responsible for the backup/restore environment, SAN and fibre infrastructure as well as installing and maintaining the servers themselves.

 

Tony Asaro: Start out by giving us some examples of how Data Domain has provided real value to your company.

 

Rob Carlisi: Yeah, I’ve quite a few examples. Recoveries with Data Domain are much faster than our previous environment. The restores start as soon as you click a button instead of waiting for tapes to mount or waiting for tape drives to become available. The footprint of our solution went from around 6 to 8 feet of floor space down to a single rack with Data Domain. Data Domain gave us the ability also to replicate all of our backups to our DR site so we have another set of Data Domain boxes sitting at our offsite location that we replicate to every day. We also use RMAN with Oracle to backup all of our databases directly to Data Domain. Since we’ve been using that method, we’ve seen up to a 60x compression rates on those backups. And also over the next few months we’re going to implement VRanger which will backup all of our VM’s over to Data Domain.

 

Tony Asaro: Rob, tell us how Data Domain has improved the overall economic effectiveness within your IT operations?

 

Rob Carlisi: Data Domain has enabled us to save exponentially actually on the amount of tape we would have had to purchase if it wasn’t Data Domain and de-duplication. We also went from a 10 to 15 day retention policy on all of our backups to a 30 to 40 day retention.

 

Tony Asaro: How has implementing Data Domain changed your usage of tape now and in the future?

 

Rob Carlisi: Well, tape no longer exists in our environment and we’re confident that it’s going to stay that way.

 

Tony Asaro: Rob, tell us about how you’ve leveraged Data Domain for offsite Disaster Recovery.

 

Rob Carlisi: Before we had implemented Data Domain, we weren’t replicating any of our data off site. When we decided to start replicating with Data Domain, we did it on a collection basis so anything that touches the primary box, gets replicated to the destination box on our off site location. We set it up the first day to replicate and it’s pretty much been hands off since then. We’ve had the remote site down. When we bring it back up, it just makes that connection automatically and starts replicating again. We’ve never had any lag and we’ve never had a problem talking to the remote site.

 

This solution actually got all of our backups from our entire environment off site. We’re able to restore at a DR site back to full throttle.

 

Tony Asaro: Data Domain’s data dedupe actually enabled your disaster recovery strategy because it reduces the bandwidth costs.

 

Rob Carlisi: Actually that’s true. After it does its de-duplication and compression and clean up, you’re not actually transferring that much data across the line compared to what it actually backs up every night. It saves on money as well as time to replicate because when you’re not moving that much data across the LAN or the WAN it doesn’t take that long to do you’re not too far behind your production environment.

 

Tony Asaro: Why didn’t you use primary storage replication versus Data Domain to do this?

 

Rob Carlisi: Well, one was the cost and we also liked an appliance environment. We found that it was a lot easier to expand a Data Domain environment rather that a primary storage environment and we just like the flexibility that Data Domain gave us. Also, the primary reason was the de-duplication technology that they had was far superior than anything else.

The Changing Landscape of Disaster Recovery

Disaster recovery (DR) means different things to different people.  For many IT professionals DR is associated with tape recovery and/or remote replication.    Recovery data from tape is fraught with challenges including rapid recovery and reliability.  Remote replication within primary storage is often complicated, expensive and inefficient requiring volumes to be restored versus specific files or data sets.

The combination of disk-to-disk (D2D) backup, data dedupe and replication offers a valuable approach to DR combining the granularity of recovering from a backup repository and the ability to leverage remote replication to store copies of data at an offsite location in the event of of site disasters and for offsite vaulting for long term retention.  For the vast majority of data this is a smart approach that serves the widest range of requirements for customers.  There will always be the need for primary remote replication but only for the corner cases in terms of who uses it and what it is used for.  However, I contend that D2D backup + data dedupe + remote replication has a much broader audience. 

Data Domain just announced an update to it's Replicator software.  I remember when Data Domain first came out with Replicator and speaking with customers that literally felt liberated by leveraging dedupe over wide area networks.  Dedupe enabled them to implement a DR strategy above and beyond what they had prior and still fit within their budgets. Five years later the solution is more mature, advanced and field proven.  Additionally, the use cases and value proposition have advanced and are proven.  Data Domain providing a combination of D2D backup + data dedupe + remote replication has impacted the data center in a number of important ways:

• Offers a cost effective and efficient DR solution that is otherwise impractical for many customers.

• Results in cost reductions include bandwidth, primary storage investment, power, cooling and floor space.
• This approach is "easy" versus perceived complications with using other DR methods (e.g. primary storage replication).
• Offers a more granular and rapid method of data recovery improving SLAs. 
• Makes it possible for customers to actually eliminate or greatly diminish the use of tape and tape libraries.  This is an important point and we are seeing momentum for tape obsolescence with a growing number of customers. 
• Multi-site replication provides a powerful and valuable solution for Enterprises with a large number of remote sites - this is enabling a new constituency to reap the rewards of this solution. Data Domain's recent expansion of 90 remote sites to a single DD690 is significant and makes it even more useful to large Enterprises. I will add, regardless of whether you have two or 90 remote sites fan into a single DD appliance at the central site is powerful especially when coupled with global dedupe. 

I have spoken to companies that did not implement remote replication because of cost - even though it was considered every year for five, eight and ten years.  And there are other companies that only replicate 10% to 20% of their data and use backup to protect the rest - they just could not rationalize the expense.  However, D2D backup + dedupe vastly changes the economics of remote replication and DR.  The landscape of DR is changing - enabled by D2D backup with dedupe - making it practical and valuable for the mainstream market.  Remote replication will drive the adoption of D2D backup with dedupe and D2D backup with dedupe will drive the adoption or remote replication.  This is a killer app on top of a killer app.