Ok, quick: How many files do you have in Primary storage? Alright, I’ll make it easier. How many terabytes of Primary storage do you have…? Give up?
If you’re like most people, you don’t know. Don’t feel bad, you’re in good company. (Gartner says that 4 out of 5 people have no idea…)
One of the things many companies I talk to are concerned about is “vendor lock-in” – buying a proprietary product or technology that leaves them with no alternatives down the road. Avoiding vendor lock-in is one of the reasons that standards exist, and certain standards are very important to everyone’s long-term future.
Over the years, as I worked with various organizations, I have detected a key trend that I think delivers a critical insight: I find that people who are open to have their perspective changed are able to adapt to our changing technology world much better than those who are not open to changing their mind. Most people listen only to information that supports their current views. This is intellectually lazy and a sure road to obsolescence in any fast moving environment. Being open grants an ability to change perspective in light of new forces or discoveries.
As tiering becomes more and more the subject of conversation, organizations are discovering that they have a lot more stale files than they thought. A good rule of thumb, if you have only one tier of storage, is that about 2/3rds of your files are stale and should be put somewhere else.
In all this talk about active files on Primary Storage and inactive files on Tier 2, rarely do I hear the follow-on discussion that I expect given the implications of what it takes to make an active file “active”. Instead, what I hear is lots of talk about taking half or more of primary storage and moving it somewhere else. And yes, this is the initial act in any tiering implementation, but it only happens once. Assuming you implement tiering correctly, you’ll never have to do this again – which also means you’ll never get this same benefit again.
The solution to weight loss isn’t exercise or eating right. Just throw out your scale and buy bigger pants.
Let me tell you a story of a ruggedly handsome technologist and friend of mine, who recently went to his doctor for a routine checkup. Now this “friend” hears the same thing from his doctor every time he visits:
As this is being written, the Industry has aligned to put all of us in the position of needing to migrate our file data from whatever system it may be on to a new platform. If your preference is EMC, then your future is ATMOS or Isilon. If you’re in the NetApp camp, then you have Clustered OnTap (8.2 Cluster Mode), e-Series or StorageGRID. Even Microsoft has come to the table with ReFS… not to mention, SWIFT, the “cloud”, etc.
Certainly a lot to think about. As you contemplate your migration, here are a few more things to consider:
One of the challenges faced by people who work over time in any particular domain (in this case our domain file data management) is recognizing the effect of steady incremental changes. These changes become like the ticking of a clock… they get filtered out of our consciousness.
Today, most organizations treat all file data as if it were equal. It is all clustered, replicated, backed-up, archived – and more… At any given point in time, the average organization has more than 12 copies of their file data. One of our customers, by virtue of their geographies and process normally has 28. (Our current record holder.)
For years we've lived with "knowledge" that NFS is faster than CIFS... While this may still be true in general, it is not true on Microsoft systems running Windows 2012 with SMB 3.0. The new SMB 3 seamless supports multiple NICs and direct-to-memory transfers making it faster than almost all other protocols. Big Data... watch out. Microsoft is coming for you!