One of the most active Cloud-based discussion topics today (which evokes widely divided opinions) relates to Cloud Backup, Recovery, Archive and Disaster Recovery. The debate quickly gets to a point where disk is positioned as an alternative to tape and therefore disk “in the Cloud” is positioned as an alternative to offsite tape.
I have been an interested observer of the debate for a long time now but believe that the time has come to speak up. I think Chris Mellor (http://www.theregister.co.uk/2012/09/06/tape_market_chart/ ) sums up the situation for tape Vs disk in a broad sense when he says:
“The tape market in general is declining because disk-to-disk backup is faster than tape, particularly in restoration, and deduplication makes drives affordable enough even though they are more expensive generally than tape. The time taken to restore from backups is more important than cash saved by not using disk”
Many Service Providers that I interact with provide proof that Cloud-based backup and recovery offerings are growing in popularity. Not only is this market more active than ever but also Service Providers are actually creating profitable businesses and providing their customers with feature-rich catalogs of variations on the standard backup/restore/archive theme.
The big question is “can a Public Cloud offering meet the needs of the broad market?” In the past the answer has been “no”, but it has evolved to “yes” very quickly. Most Public Cloud offerings provided an attractive option for backup/restore/archive to a narrow market. The challenge for most Service Providers has been how to offer a backup/restore/archive service built on a particular technology stack to a customer who has their primary data residing on a different technology stack.
Let’s look at an example of a Service Provider who builds their backup/restore/archive offerings on a NetApp infrastructure. They are providing a specific “target” environment for customers to send their data to. If the customer has their primary data stored on a NetApp infrastructure (the source) then the required features (in addition to the transport mechanism) would be handled by NetApp’s native software layer. Fairly straightforward stuff and the basis of numerous Service Provider offerings today. Like for like.
The situation gets a little more interesting when the source is a different technology from the target. Like for like simplicity disappears. No problem. This is where specific technology partners can step in and add a layer of value to the solution. The most active technology innovators I see in the Public Cloud market are Asigra and Commvault (but the list is growing). Both of these organizations enable the Service Provider to add breadth to the type of customer environments they can serve with their NetApp-based Cloud infrastructure.
Now the debate related to disk (in the Cloud) versus tape (in the warehouse) becomes more interesting. For the Public Cloud to offer a real alternative to offsite tape services it had to reach a point of maturity where the statement “anything you can backup to tape you can backup to the Cloud” was both valid and accurate. That point has now been reached. But is that the end point as far as the debate goes? I don’t think so.
The debate quickly evolves into one of cost. There are many studies and models that show financial justifications why offsite tape is a cheaper option than disk-based Clouds. There are also studies that show the reverse is true. I believe this aspect of the debate is missing the point.
The fundamental reason why disk-based Clouds are better value to a business than offsite tape solutions is related to the changing nature of data and the value that is placed on it.
Data has evolved from a focus on keep to manage to leverage. This is why disk-based Public Clouds make sense and, more importantly, provide real and quantifiable value to businesses.
Historically, businesses were encouraged to keep their data. There were numerous regulations, best practices and industry standards that determined how much and for how long a particular business should to keep their data but depending on what industry the business operated in, the penalties for not keeping old data were relatively minor. This was the environment where tape provided a viable solution to meet the needs of the business and ensure that non-current data was available if required. It enabled a business to “tick the box” in relation to data retention.
As the volume of data a business generated grew and the pressure to keep more of it for longer periods grew the focus shifted to the management of the data. This is where offsite tape services provided real value. Not only was the non-current data being kept but also the likelihood that it could be restored if required was high because it was being managed by professionals who could meet SLAs.
Today the role of data in a business sense has changed dramatically. Non-current data has a value that is being recognized and exploited (when possible). In some cases non-current data has as much value as current data. Analytics has changed the way businesses view their data, both current and non-current. Data mining is now a necessity not a luxury that only the largest and wealthiest businesses can afford to undertake. This is where the debate of tape versus disk needs to move from a focus on cost to a focus on business value.
Service Providers who offer Public Cloud services which store (keep) and manage non-current data for customers are now in a position to offer a portfolio of ways to leverage the data. Many Service Providers are doing it already but there is still more innovation to come in this space.
So when you read another installment in the long running debate about tape versus disk ask yourself if the argument addresses the ability of the data owner to leverage their data assets. My guess is that the argument will quickly shift from a Private Cloud (disk-based) to a Public Cloud (also disk-based) one and tape will disappear from the vernacular. A little bit like it has when I think about my music (assets)……………..no room for tape in that paradigm.