There’s less milk than meets the eye using Tegile’s dedupe and compression

People have a strange way of linking everyday items and events with seemingly unrelated aspects in life. Some would call these analogies, or comparisons between two things. I just call it concept dyslexia.

My most recent one happened the other day in the shower — yeah some people sing — I just think “outside the shampoo bottle.”

I was thinking about Tegile’s inline dedupe and compression and couldn’t help but think about those trick milk glasses that magicians use to make it seem like there’s more or less liquid in a glass than there actually is. And of course there’s a $20 dollar bill in the glass that miraculously appears after the liquid vanishes.

When you see the trick it leaves you wondering how they did it. Sometimes people respond that same way to the unbelievable gains in performance and capacity that come from using Tegile’s dedupe and compression.

Without revealing how the magic milk trick works I can tell you how Tegile’s inline deduplication and compression works.

When a write comes in, we compress it immediately and then take a hash on it. With Tegile’s patented accelerated metadata caching in which we use high-speed SSDs to store metadata, we quickly look up that hash to see if it’s already been written. If it has, we simply update the metadata. This gives way to unprecedented performance and capacity. Knowing the secret to this ‘trick’ may not wow your co-workers in the breakroom — but your customers will definitely feel the difference (as will your budget).

So now that we know the trick about dedupe and compression (don’t tell anyone), how does that help your storage infrastructure? The first thing I think of are those ubiquitous shampoo bottles scattered around the shower and how they remind me of all those desktops in a VDI environment.

It would sure be nice to dedupe those shampoo bottles. Luckily in a VDI environment your desktops get deduped. Check out the screenshot below of Tegile’s famous ‘donuts’ that represent the size of original, deduped and compressed data. The below data donut is taken from a recent Tegile customer using VMware’s ESXi. (Also look at our 1000 seat VDI reference architecture).

Tegile’s Famous Donuts

In the above screenshot, you can see that the original data was almost 44 TB. But after deduplication and compression there was only 7.66 TB written to disk. This means they are saving over 82% of their storage space! This is that exact scenario where the milk disappears and $20 pops out. However, here the cost savings to an IT organization is much more substantial than $20.

A VDI environment is deduplication’s sweet spot. But what happens if you’re running an SQL environment? Databases don’t really have data that dedupes well but no worries though, you can easily turn on and off deduplication and compression in the Tegile IntelliFlash array, as shown in the below screenshot.

Project configuration — Dedupe and Compression Settings

Changing the settings is a simple task of pulling down the menu. In most cases lz4 is the compression to pick and is chosen by default when creating a project; and in the case of VDI, deduplication being on is a must. But let’s say you had created a LUN in a project that had dedupe on and you realized at some point that you really wanted dedupe off. That’s no big deal here because changing the settings on the fly is easy. You simply edit that specific LUN’s settings and turn dedupe off. Now all subsequent data will no longer be deduped. Meanwhile all the other LUNs in the project will continue to have their data deduped.

As you can see, the features of deduplication and compression can be beneficial in specific environments and can provide immense performance increases and storage capacity savings.

With the flexibility to turn this functionality on and off, you can quickly adjust your array to perform at peak levels for each individual environment, which reminds of those shampoo commercials where they say the shampoo adjusts to your hair’s daily needs! With Tegile IntelliFlash arrays, it’s adjustable to your needs — and I bet you’ll find there’s less milk than meets the eye using our dedupe and compression.

2 comments on “There’s less milk than meets the eye using Tegile’s dedupe and compression

  1. Kris on

    When Tegile arrays dedupe the data, do they check for matching hashes against those across the entire array lor do they check for matching hashes only against metadata of that specific volume?

    • Harrison Waller | Technical Marketing Engineer on

      Great question! Our dudupe table is per pool (which contain projects and within that LUNs/shares). But you can turn dedupe on or off at the LUN/share level.


Leave a Reply

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