The following is an email that I received from Zsolt Kerekes, Editor of STORAGEsearch.com. If you have not checked out his blog, it is very informative. I stumbled upon it because I wanted to take a closer look at the storage sector. As a former media banker, storage was like potatoes or hot sauce to me – who knew there were so many different types?!
The following is an email excerpt from Zsolt that I would like to share. He provides some insights regarding Big Data.
___
John thanks for your comments and question.
re – big data or is it big hype?
It can be real or it can be hype.
True big data architecture (for SSD) starts at the internal level of the design.
How many memory chips has it been optimized for? Once that decision is made – the cost, space, reliability and performance characteristics are inevitable. My article – http://www.storagesearch.com/bigvsmall-ssd-arch.html may be too technical for you – but here’s an analogy.
If you look at the I/O of food in your local supermarket. The way it comes in from suppliers (from trucks) is different to the way it goes out (in cars).
Technically the supermarket could receive all its supplies in small car loads – but that would be inefficient.
Technically your house could receive all its supplies in big trucks (if the road permits) – but that would be inefficient too.
There are market flexibility advantages which come from designing small architecture SSDs . You can use the same design in a notebook or a server so you can access a bigger market.
But big architecture SSDs intrinsically can work with 30% to 40% less memory chips – and use a wider range of reliability / data healing tricks.
On the downside – big architecture designs can’t be used in as many markets as small ones – so their cost base may be higher.
The result is that the purchase cost advantages of one versus the other may cancel out – as seen in street prices of arrays.
Running cost – if you have a big system – is something which remains.
===
When it comes to hard drives – RAID was better than what it replaced in the 1980s – but if you started again today with the internet connectivity and processors we have today – and started to design big arrays of disks from first principles then you wouldn’t see just the RAID systems you see today. That’s because RAID is small architecture too. It optimizes over maybe 5 to 10 disks. If you optimize reliability etc over 100 disks – then you get better efficiency. That’s why Google designed its own disk managment system. The cost savings at the million plus disk level are worthwhile.
This is exactly the same principle in SSDs. SandForce optimize around less than 10 flash chips – whereas Violin optimize around 100 plus.
===
Technically you can wrap “big software” around either type of hardware architecture to simplify “big data management”. But the electricity cost isn’t fooled by the software.
Feel free to share my reply if you find it useful.
It was more interesting than what I was supposed to be working on.
Take care
Zsolt
From: John Doyle
Date: Thursday, March 08, 2012 19:43
To: Zsolt@STORAGEsearch.com
Subject: Good Stuff
Zsolt,
I just wanted to send you a note saying that your website is very informative! Thank you for putting it out. Also, let me know if you have conversations with bankers.
I’m a tech banker and storage is becoming a very hot sector. I’d like to get your thoughts on whether you believe in big data or is it big hype.
Hope all is well and thank you!
Best,
John
John H. Doyle II
Managing Director & Founder
Peachtree Capital Advisors, Inc.
1055 East Colorado Blvd., Suite 500
Pasadena, CA 91106
OFFICE 626.204.4047
MOBILE 626.394.2167
FAX 626.628.0411
Investment Bankers
Software, Digital Media & Cleantech
www.PeachtreeCapitalAdvisors.com