Don’t be Duped by Dedupe Claims

Donuts to the rescue!

In enterprise storage, we are in the early stages of a major revolution. This revolution is driven by the emergence and adoption of technologies such as flash storage, online data compression, object storage and software-defined storage.

See Gartner’s graph of the “Hype Cycle” for various storage technologies. Note that solid-state arrays are approaching the “Trough of Disillusionment.”

Given the fever pitch of flash over the past few years, what contributed to the rapid descent of solid-state storage into disillusionment? I’d venture to say this disillusionment is mainly attributed to the realized economics of effective usable capacity through deduplication.

It’s a given that solid-state arrays perform an order of magnitude better than legacy enterprise storage. And with solid-state arrays, companies can achieve tremendous I/O acceleration across a broad range of business applications. However, companies have also seen a wide range of economics ($/GB) that stem from the over-inflated claims from storage vendors of efficiency in data reduction caused by deduplication.

Too many companies have been duped by dedupe ratios.

So then, how then do you estimate the real-life reduction ratios that you are likely to achieve with your data sets?

Donuts to the rescue!

At Tegile, we have taken great effort to assimilate information across thousands of Tegile Intelligent Flash storage arrays (all-flash and hybrid) deployed in customer environments. This data analysis allows us to characterize the extent of data reduction achieved across a wide range of enterprise applications. As part of the IntelliFlash Management console, each Tegile storage array provides charts that show the extent of data reduction that a user achieves across storage pools. We affectionately refer to these charts as “Donut” charts.

Enjoy the donuts!

See these examples of Donut charts that several hundreds of our customers have sent to us as part of our program, Donuts for Donuts. These Donut charts are categorized by application type and configuration. (To maintain confidentiality, customer names are not disclosed.)

Use below charts as guideposts to estimate the kind of data reduction that you are likely to achieve.

Virtual Server Infrastructure

Web Server + Biz Apps + DBs

Test / Dev Apps

Financial Applications

Desktop Virtualization

VMware View on vSphere

XenDesktop on vSphere

VMware View with Linked Clones

Virtualized Databases

Test / Dev MS SQL

Virtualized Oracle

SQL Analytics on vSphere

Mixed Application Workloads

Cloud Provisioned VMs

DBs + File Services + Sharepoint

VMs + MS SQL + MS Exchange

When evaluating storage vendors

As you evaluate storage vendors, ask for proof and a guarantee. Ask for proof of data reduction achieved by each vendor’s customers. And ask if your vendor is willing to provide some form of guarantee with respect to data reduction.

Stay tuned

Next week we announce the winner of Donuts for Donuts! Stay tuned to learn which Tegile customer had the most data compression — and will win donuts for a year!

Leave a Reply

Your email address will not be published. Required fields are marked *