Disaster Recovery Acceleration Essentials: What you need to know

Grab your FREE copy of the WAN Optimization 2.0 White Paper

Hello, and welcome to another Quick Bits session on Riverbed technology and optimized disaster recovery. This session will cover how Riverbed significantly accelerates disaster-recovery operations so that organizations can speed up jobs like data replication or WAN-based backup to protect their data more efficiently and increase the resiliency of their IT infrastructure. These are some of the highlights of how Riverbed can actually help organizations enhance their overall DR strategy.

Reducing the RTO and RPO for an organization. We dramatically reduce the amount of time required to back up or recover data in the face of a failure. So, literally, taking a backup window that was taking 24 hours or more, an IT operational impossibility, and reducing that down to four hours or less, or taking a NAS or SAN replication job and reducing that by 75 percent, and going from eight hours down to one or two hours to actually complete that transfer.

And ultimately, this lets organizations snapshot data more frequently so that they can reduce their data-loss risks from corruption and actually increase the number of data snapshot points that they’re actually taking throughout the day.

Overall data protection. We’ve got financial organizations all over the world who are consolidating their backup infrastructure and eliminating tape media from their remote branches. This is helping protect organizations or reduce the risk of becoming a headline story about a tape falling off of a truck somewhere.

And ultimately, saving money from bandwidth utilization, by reducing the need to install massive pipes into their DR sites, using Riverbed technology.

And last but not least, virtualization–things like VMware playing a huge role in business continuity and disaster recovery. We’ve got several organizations who are using our technology to essentially help replicate VMware Server images across the WAN to a DR site.

So the question there becomes: what is the business value to your organization to have your entire server infrastructure replicated on a daily basis and be able to fail over to your DR site with no performance hit? And that’s the type of solution that Riverbed Steelhead appliances can enable.

Many of you are probably already familiar with this slide. The point being here that, with regard to disaster-recovery-related operations, the data-streamlining and transport-streamlining capabilities and RiOS are really what are coming into play here. So, the ability to de-dupe data and reduce the transfer of redundant data over the WAN, as well as perform a virtual-window expansion to eliminate TCP chattiness and latency over the wide-area network, are really what are coming into play in the disaster-recovery arena.

This is just a sample topology of how a customer might implement Steelhead products in their environment today. And as I mentioned earlier, a lot of organizations are already using us for things like WAN-based backup.

And an important thing to note here is that we work, out of the box, with most of the major backup solutions on the market today, because we’re a network-based solution, so we perform our data-deduplication and transport-layer optimization on data across the WAN, completely independent of the backup software. As a result, we work with most of the major vendors, like NetBackup, Backup Exec, Tivoli Storage Manager, CommVault, CA, Legato, if I haven’t already mentioned that. A host of others–Double-Take, for example.

So, in general, organizations will see a five to 50X acceleration, depending on the vendor that they’re using as well as the working data set.

And a lot of organizations are also using us to enhance their data-replication strategy to a remote DR site. And as a result, we also partner very heavily with most of the storage vendors in the industry. So, vendors like NetApp, for their SnapMirror, SnapVault, as well as ReplicatorX products; and organizations such as LeftHand, Isilon, and EqualLogic as well, as well as Hitachi Data Systems, too. So, a number of different storage vendors, where we’re working alongside them to help their data-replication capabilities.

And what Riverbed brings to the table that’s a little bit different than other WDS vendors in this space is that we have this behavioral traffic-recognition capability, to essentially recognize large-scale data transfers in flight. And we kick our appliances into another gear to essentially maximize throughput over the WAN. And this is what generally makes organizations get better results with us on large-scale data transfers across the wide-area network.

Another way Riverbed can help organizations with their DR processes is essentially enabling a fill-the-pipe strategy for high-bandwidth links over a long distance. And essentially what this does is, for organizations who’ve got a DS3 or an OC3 or higher over a long distance, the inherent limitations of the TCP protocol basically only allow an organization to get 20 to 30 percent utilization over that long-distance pipe. And what Riverbed can do is help enable a faster TCP ramp-up, to get better throughput over this type of high-bandwidth link.

This is just an example chart of the differences between the different TCP mechanisms that we employ on a Steelhead appliance. You’ll see that the normal TCP ramp-up over a high-bandwidth link will only allow a certain amount of throughput. Whereas, with max-speed TCP or high-speed TCP, both of which are available with a Steelhead appliance, you can basically ramp up TCP further and essentially police the bandwidth throughput on that high-bandwidth link.

And one of the things that really makes Riverbed unique is our disaster-recovery acceleration capability. And this assumes that you already kind of have an understanding of the data-streamlining and transport-streamlining capabilities of the Steelhead appliances. So, if you don’t, you may want to review that, just to get a better sense of what’s going on here.

But essentially, what happens is that we recognize a large-scale data-replication or backup job. In many cases, this is a multi-terabyte job. We will enable our DR acceleration and essentially kick into a different gear on the appliance to maximize throughput.

What this does is we enhance our disk utilization by writing and reading data in a more contiguous fashion on the disk. And this is basically in a way to try and overcome the laws of physics by minimizing the amount of time that we’re actually randomly accessing a disk’s spindle in the appliance. And this will ultimately have the effect of increasing throughput over the WAN so we’re able to replicate data faster.

In addition to that, we also have the ability to shift between algorithms, to basically optimize our data-reduction capabilities in terms of how we’re actually pushing data through the pipe.

So this next session will just cover some of the customer results that people are seeing using our Steelhead appliances in conjunction with disaster recovery.

At a high level, here are some quotes from customers, in terms of some of the results that they were seeing. Mercury was using NetApp filers and SnapMirror to replicate their data. It was taking 24 hours to do that. They cut that down to six to eight hours.

GeoEngineers was doing a DFS SAN replication using EqualLogic. They went from over two hours down to about 10 minutes, I believe.

And Liz Claiborne was using CommVault to do WAN-based backups. I believe it was CommVault. Going from 16 hours to around six hours.

And Cubist Pharmaceutical, heavily virtualized environment, replicating all their VMware server images to a DR site in minutes, daily, and seeing a 50 to 55X improvement in the replication times.

This is a case study on GeoEngineers. It’s available for download on our website as well. And they were using EqualLogic as well as Riverbed to perform a Microsoft VFS replication back to the SAN, as well as a SAN-to-SAN replication for DR as well. So, a three to four X increase in WAN capacity, and reduced their data-replication times down from two hours to 10 minutes.

Cubist Pharmaceuticals. We talked about this earlier. A 90-percent virtualized environment, using VMware. They were using VMware Consolidated Backup, VCB, to replicate VMware Server snapshots to their DR facility over a one-meg link. They were able to replicate their entire virtualized server infrastructure on a daily basis, and in addition to that, were replicating 120 gigs of Oracle data on a daily basis and able to complete that operation within two hours.

This is from a competitive evaluation done by a Fortune 100 company last June, and it essentially was a four-month evaluation of Riverbed, along with two other competitors. And you’ll see in the right-hand column here that for NetApp SnapMirror, Riverbed was anywhere from roughly four to 10 times better than what the competition could offer from a throughput standpoint. So, in this particular case, the bigger the bars, the more throughput, which is a better operation, from a data-transfer standpoint.

So, just to give you a benchmark standpoint, again, this was a four-month evaluation, where we were selected as the technical winner.

So, in summary, Riverbed can help accelerate a number of DR operations for your organizations. We can help speed up the replication and backup of virtual machines and data. This helps organizations meet SLA obligations for the recovery of data and operations within their business-continuity plans.

And more importantly, there’s no performance hit at the DR site. So a lot of companies us a smaller WAN-link pipe into their DR facility to help save on costs. But that site is often located halfway across the continent somewhere, and when your users fail over to it, the performance is almost unacceptable, or unusable, from their perspective, so it’s the equivalent of being down. With Riverbed Steelhead appliances, you don’t get that type of a performance hit, and you can effectively use your DR facility as if it were your primary data center in the face of a failure event.