Promises of 12:1 compression and deduplication one year out
Language can be as precise as the user intends, and that intention is important to understand.
Consider two words, very similar but not the same: “shall” and “should.”
“Shall” can be defined as “going to be,” “definitely,” or “will.”
“Should,” however, means “ought to,” “expected,” or “probable.”
Why is this exercise in the English language important? Because “should” is not the same as “expected.”
Case in point: data storage
The latest popular feature is compression and deduplication – appearing as the workaround for offsetting the relative expense of Solid State Disk (SSD). Array vendors are communicating that you really do not need to purchase the full capacity to fit the requirements because compression and deduplication will enable a much smaller footprint. Great theory? A lot of marketing claims are being made. One claims 30% to 75% capacity savings, another a 5-10x reduction. Yet another specifically touts 75%. The first two are interesting just because the range is so wide that it looks more like a disclaimer than a fact. Guaranteed?
I am sure that the compression and deduplication technology works—it is not my intention to deride it.
I am also not claiming that one vendor’s technology is any better than any other’s.
Your concern should rest around the “should.” If a buyer is led to believe that the storage solution should provide a 12:1 compression and deduplication and, after being in production a year or two, the solution yields only 5:1, what are the options after the purchase? Is that storage manufacturer going to make up the difference in capacity utilization at no cost? Doubtful—you bought it, you own it – they’ll be happy to sell you more.
Future-proofing is also another consideration. What if, as an example, a hypervisor provider introduces compression and deduplication natively into the OS? Then what?
Claims of compression and deduplications are a great marketing play, but smart buyers recognize that data changes over time. What will your data look like in three years?
Predictability is the key. Look for a solution that “shall” provide full, rated performance at maximum rated utilization. Uncertainty is something you do not want in your infrastructure; 20TBs today is 20TBs five years from now.
Want the facts on how to achieve 100% performance at 100% capacity? Read on…