Comments on: Ellison Takes Aim at HANA, But Misses the Big (Data) Picture

By: Rob Commins

Rob Commins — Thu, 04 Oct 2012 15:46:00 +0000

My apologies if I came across as an informercial for our wares. My intent is to be very up front about the position I was writing from. I do recognize that my comment is a bit tangential, but we see so many vendors making claims (as Ellison did vs SAP) that have little or nothing to do with what a customer will see in production, I felt compelled to post my comment with anecdotal examples I know from my day to day business. We do indeed have reference architectures and best practice guidance we’ve assembled from customer’s real world production feedback. If your schedule does not permit your own POC, I suggest customers looking at new technology press vendors for those real world examples and leverage references from end users to validate those claims. Hope that helps.

By: bugoff1234

bugoff1234 — Thu, 04 Oct 2012 14:41:00 +0000

Now this is an weird comment… are you advertising something Rob?

I do commend you for asking for a POC, but many other of my IT administrator peers would agree I don’t always have time for POCs. I’ve got my day job to do. Do you guys have validated designs?

By: Rob Commins

Rob Commins — Wed, 03 Oct 2012 22:56:00 +0000

Hi Jeff –

Great coverage of the great in-memory debate. I am with a new data storage firm, Tegile Systems. While we are not the juggernaut that Oracle is, we often get pulled into “mine is bigger than yours” stand-offs. Here’s the thing: we are big on de-duplication and compression here at Tegile. While intuitively, this helps customers save capacity on the rotating disk at the back-end of our arrays, it actually has a huge positive performance impact to applications such as databases. How? Our data reduction algorithms also save space in our DRAM and SSD space too – we use these as large read and write cache pools, respectively. So, if I can get a 5X data reduction ratio, I can also jam 5X mode data into cache, thus boosting cache hit ratios and drive performance much higher than what I might normally get with a traditional storage system.

Why bring this up on your post about Oracle and SAP? Only to warn readers that apples to apples comparisons are even harder to assemble today than they ever have been before. Simply looking at the cache capacity of our mid-range arrays that have between 600GB to 4.4TB of SSD in them doesn’t do the architecture justice. Customers need to demand an in-house proof of concept to put the “your mileage may vary” weasel words to rest. Arm waving during the keynote of a conference means little to nothing (and I’m a marketing guy!)

Again, thank you for the great coverage, Jeff.

Rob