Tony's blog

Federal Forum with Data Domain Customers

Last week I went down to Washington DC to present to a number of Federal customers on various industry topics. I asked them a bunch of questions and here are the highlights:

- There was one deployment of FCoE and everyone else was highly interested in FCoE going forward. I mentioned the political challenges of FCoE between the LAN networking guys and the SAN networking guys -- and the guy who actually deployed it said that he found no issue. The lines are still there and remain - the SAN guys still manage FCoE. The others in the room were still unsure how this was going to all play out. Again, no one in this group deployed iSCSI. I am a big proponent of iSCSI but FC still rules the SAN world in larger organizations.

- Everyone was using VMWare and no one had deployed or was overly interested in Hyper-V. This is consistent with other forums I have been involved in recently.

- There was some interest in SSD or EFD for Enterprise storage, but only one person has deployed. His take was that it was a waste of time and money. It did not give him the performance he was looking for. This is an important point. Storage performance is complicated and depending on the I/O profile, SSD/EFD may not be the answer you are looking for - it is not a panacea by any means. It would be great to remove a mechanical device from the data center - after all disk drives do three things: they read, write and break. But for the most part disk drives still provide the best price/performance and even with the advances in SSD/EFD, it still is only right for certain corner cases.

- There was no implementation of active archive storage with this group. There was a level of interest but they agreed that the challenges were identifying what to move, how to move it transparently, what to move it to and ensuring response times if that data is ever needed again. One person piped in saying that there was no single solution that allowed them to do database, email and file archiving without putting agents on their hosts. I suggested to go after the biggest problem first - which is unstructured data. That could save them a ton of money. There were nods in the room but you could tell that they weren't running out the door to make it happen.

- There were no plans for any of these folks to use a public cloud service within the next twelve months. And by the looks of it there wasn't any real interest beyond that timeframe at this point. There was only one company that was building their own private cloud. I wanted to pick their brains more about what they specifically were doing but they had to leave before the meet and greet session (I hope it wasn't my presentation that made them leave).

- Finally, it was good to see that close to half of the IT professionals had Green IT initiatives. Naturally IT folks have been focused on power, cooling and floor space for years and now a number of them were going green. While this is related it is different - having a green mandate goes beyond these three things (although it is a good place to start).

D2D Backup + Active Archive

In my travels I have discussed the value of storage tiering and it is amazing how few companies have actually implemented this. Part of the challenge is identifying what to move and when. One way to easily address this is to just consider dormant data. Over the last few years I have presented to hundreds of IT professionals and the vast majority believe that more than 70% of their data is dormant 90 days after its creation. If this is the case, then focusing on dormant data after a 90 day period seems like a good place to start - move this data from your primary storage systems to a lower tier.

The next challenge with storage tiering is identifying what solution you are going to use to move the dormant data from primary to archival storage. This process needs to be transparent to the users and applications. And it would be nice if it was easy for IT professionals to implement and there was a high level of performance. In the NAS world you can use file data movers or virtualization solutions. Data Domain partners with F5 and its Acopia solution.

The third and final macro issue is what platform should be used for the active archival storage. Data Domain is typically seen as a disk-to-disk backup solution but it should also be considered as an active archive as well. Remember that the Data Domain solution is a NAS system and therefore a mount point can be easily created and it can be used to store files via NFS and CIFS protocols. One of the main reasons to do this is to take advantage of its dedupe capabilities. Since this is not a backup data - the dedupe rates will not be as high but you should easily get a 2-to-1 and maybe as high as a 5-to-1 ratio. What makes this even more compelling and valuable is using Data Domain for both D2D backup and as an active archive. It is this combination that presents a powerful value proposition. Consolidating these functions into a single Data Domain appliance further reduces cost and complexity. In addition to the direct CAPEX savings there is also the reduction of power, cooling and floor space.

There are companies that have saved significant capital costs by implemented tiered storage environments - even without dedupe. Imagine adding dedupe to the equation. Further, leveraging the same applicance that is being used for D2D backup and giving you 20-to-1 data dedupe. Not many people talk about providing a consolidated D2D backup and active archive solution, but it really does make a ton of sense if that single solution can support both with dedupe.

Technology + Product + Business Excellence

There is some discussion around the notion of data dedupe as a feature or a technology and that it will some day be pervasive. People having been saying this for more than five years and it is still not a reality. There are actually only a few vendors that actually have successful solutions in the market supporting this valuable but complex technology. The fact of the matter is that dedupe is complicated - creating dedupe algorithms are only a part of the challenge. Data integrity, performance and scalability are key considerations. If dedupe was easy then it would be pervasive and it is not.  

 

It is then critical to manifest dedupe in a product that has been designed from the ground up to support the specific requirements for its use. In the case of Data Domain special attention was paid to the fact that dedupe is CPU-intensive. Additionally, they chose to support NAS for ease of use and transparency. They also focused on the file system and adding technology that ensured data integrity.  There are dozens and perhaps hundreds of big and small considerations to be found in this process.  There are other things to consider as well that live outside of the appliance. In speaking with a number of Data Domain customers I've heard more than once how good their documentation is. I have rarely heard this type of praise for any IT vendor.  

 

Business execution is critical to the process that includes a number of things. It involves putting together a strong team, a powerful and consumable message, an effective and scalable go-to-market strategy, a strong sales and support organization, etc, etc. It is also important that there be strong business fundamentals - meaning that having a product that makes a really good profit for those who sell it while at the same time saves money for those who buy it. And customers should care about this as well because they want healthy vendors to work with that will continue to support, innovate and execute for the long term.  

 

It is the combination of Technology + Product + Business execution that results in leadership. The people that discuss dedupe as just a "feature" or a "technology" I believe miss the point - as evidenced by the success of Data Domain.