Dell, with their announced acquisition of Ocarina Networks, put
themselves squarely in the deduplication market joining both NetApp,
Oracle (SUN) and EMC as storage system companies offering some form of
primary storage deduplication. While it remains to be seen what Dell
does with the Ocarina technology, it clearly sends a signal to other
storage system vendors that primary deduplication is going to be a
required capability just like snapshots are today.
Dell could do a few things with the Ocarina product. It is safe to
assume that they will use the current Ocarina product as is to offer
deduplication to their entire customer base. The Ocarina product was
originally sold as an appliance that can be added to nearly any existing
storage infrastructure. It seems reasonable to assume that Dell will
also begin to look at integrating the technology into their EqualLogic
family of products. It will be interesting to see if and how Dell uses
the Ocarina product with the products that it OEMs from EMC. They could
use it to differentiate themselves from the throng of EMC resellers. All of that is conjecture right now, and we will have to wait and see
what Dell does. More interesting for the moment is what exactly does the
rest of the storage community do?
Clearly Dell joins the growing list of
storage systems manufacturers endorsing deduplication of primary
storage as an important element in the product offering. The challenge
that the remaining storage manufacturers face is that if they don't have
something already ready to implement, they are running out of time to
develop their own. Unless they have a deduplication development effort
underway, there are two choices: buy a deduplication technology or OEM
one. From the buying perspective there are fewer companies left to purchase.
While it is safe to assume that another company may come along to offer
deduplication as an OEM service, clearly the choices are limited and
becoming more so. Also as we move out of the larger tier of storage
vendors there is less money available to buy a company offering primary
storage deduplication technology.
From an OEM perspective there are products that either are file systems
or are add-ons to file systems that provide deduplication capabilities.
These vary in scope and efficiency but for file based customers they may
have an appeal. The time to market may also be faster since there is
less integration work. The obvious limitation here is that these are
file system-based solutions, and as a result the deduplication
functionality is limited to the file system that it controls. Using this
type of solution may mean a bit of a feature burn for some storage
manufacturers, turning off much of their intelligence and leveraging
what the file system gives you. Depending on the level of data services
the storage system provides will determine if this is a net gain or
loss.
The other option for storage system manufacturers is to integrate at an
API set level deduplication hardware or software. While this will take
extra development time on the part of the storage systems integrators,
each should not be a half-year project. Anecdotal evidence suggests maybe
a few weeks. The advantage of this approach is then you have a native
interaction between the storage controller and its data services and the
deduplication engine. This should mean that all the investment that the
storage system manufacturer made in providing data services like
snapshots, thin provisioning and automated tiering could continue to be
controlled by the storage system's manufacturer. It also means that the
storage system manufacturer should be able to offer deduplication across
all storage types NAS or block. These models each have their own distinct value. No matter what choice
the storage manufacturer makes, it is becoming clear that primary storage
deduplication is quickly moving from something nice-to-have category to being a must-have.
