Techno Musings

So I’ve read and poked around a bit on BackBlaze’s cheap storage, and there is a key point that I want to bring up: Total Cost of Ownership!  Om Malik at GigOm makes the following point:

Nevertheless, this move by Backblaze is interesting because it addresses the current logjam in the hardware business. At our Structure 09 conference, Facebook’s VP of engineering, Jonathan Heiliger, lamented how the chip industry and hardware makers fail to address the needs of the big spenders: web companies. Facebook had to build its own mousetraps to met its specific needs. If your startup has open-source hardware designs that meet the needs of today’s web-based businesses, you can easily do an end run around the incumbents.

Note the point that I’ve bolded above about big spenders.  Further over at c0t0s0d0 there is a good criticism of all of the buzz surround the BackBlaze device, which I think is well worth the read including the follow-up from the author of this site.  He makes excellent observations that there is more to the tier one storage vendors attention to quality and that warrants being paid for.  He also makes a great point about data integrity and what he calls the recent attitude of being able to lose data.  I agree with him here that this is a sophomoric attitude, because things like the FRCP in the USA mean that showing up with a corrupted file and lead companies to loss of brand reputation and cold hard cash from settlements or even fines from a loss.  Just imagine showing up and saying “I’m sorry your honor but we downloaded this storage spec from the Internet and as a result this cluster of files is now corrupted and we don’t have an alternative working copy.”  I can hear the judge now in my head: well that is too bad for you.  (Mind you I know that the SW layer keeps copies, but I’m intentionally exaggerating here to make a point.)

I want to leave Joerg’s points and come back to the issue raised by Jonathan above: namely big spender web companies.  These companies, and in the past and present, as well as Universities have something that most smaller companies do not: lots of software developers.  (Note that Joerg has a great analogy about Software in his follow-up which I will not repeat here, so see if you can find it as it should produce a chuckle.)  Google, Backblaze, Carbonite, Amazon, etc. all have an extraordinary number of software developers on staff who can create their own stacks that suit their needs.  Further Universities, who are also cited to be interested in this type of system because of cost, have a pool of very very very cheap labor called graduate students who can also create loads of software or manage sickly brittle systems at almost no cost — in fact they are positive cash flow to the University.  It is the people and long term costs of maintaining software that the rush to this kind of architecture is dependent upon.  In fact TCO numbers from IDC point to the fact that TCO continues to outstrip capital acquisition costs significantly whereas the capital acquisition costs are remaining largely flat.  What is going on with the various storage vendors out there when they add all of the Reliability, Availability and Serviceability (RAS) features to storage and innovation on top is that you are effectively paying them for reducing your staff.  As a consumer you are saying, hey I don’t have the unlimited budget of Google or the nearly free labor pool of a University so I will trade increased capital expenditures for a product that is far more reliable and with usability improvements that mean I can deploy fewer people to get the same job done. This is one of the basic reasons you, as a consumer of technology, are buying from trusted vendors we stick around (I should mention that Hitachi as a company turns 100 next year), we’ve proven we can add quality and make our products robust, and listen to you as our customer.  So if you want to buy cheap storage please don’t forget the hidden costs associated to TCO which Hitachi specifically helps our customers put to bed.

Now tying this back to Chapters 1 and 2 what this implies is that for those users who are in older generations and have emotional hangups which I mentioned, Hitachi can help you out here.  You most likely do not have an unlimited budget to hire developers or get them for free that can nurse along homebrew software and hardware, so you depend on Hitachi.  Rest assured Hitachi’s products are well engineered to put even the most risk adverse Baby Boomer at ease or satisfy that highly skeptical Gen-Xer that wants to know where his data is or that he can reliably be retrieved.  Note that in a future chapter, I’ll be exploring some of Hitachi’s offerings in the private cloud space today and how we can help you get past the hangups.

Tweet

Posted in Cloud, Hardware, IT Transformation, Innovation, Tech Talk

Comments (6 )

Post Comment

Sorry Michael but the Hitatchi 100yr heritage means zero in todays climate and thus a companies legacy counts for nothing when it comes to long term data retention.

Sun: gone, SGI: Gone, the biggest of banks:gone.

“Your Data Should Outlive your Vendor”

The only way to guarantee data longevity is to rely on the underlying technology being open and not vendor tied.

http://www.matrixstore.net/2008/11/07/will-your-data-outlive-your-vendor/

[...] adage “you get what you pay for” keeps entering. Michael Hay wrote an interesting blog here, where he looks at “cheap storage” and rightly points out that one of the costs that [...]

Nick, I read your blog post and while I can understand your logic, I think that you are ignorant about the Hitachi approach to digital archiving, and this is where I want to start my criticism of your logic. HCAP (Hitachi Content Archive Platform) always started with the approach of allowing our customers’ data to easily be extracted from the system. As such the persistent contract we make with the client is the file name and a durable path and persist the file in the originating format. Note that we don’t use proprietary protocols either, instead sticking to a suite of REST/HTTP, WebDAV, NFS, CIFS and SMTP. I should also mention to make it simpler for the customer to extract their data from the system we provide the archive object package format that extracts the object content, its explicit metadata and its implicit metadata. I should also mention that we push our partners who write applications that talk to HCAP to open up their data formats and get past the “lock-in by obfuscated file format syndrome”. This is actually a scary proposition for many of the ISVs, but once they do it they can move on to building more user valued IP. I should also mention that the “feeders” we have for our archiving system HNAS and HDD-MS also do not change the file format either. There are tangible benefits to us because it allows us to implement more complex value added use cases for our customers like search federation, risk scoring across a continuum of file storage, set up rich ILM policies for migration, coordinate between file and block tiering, etc.

As a vendor we pick open technologies for implementation as well. We derive our internal Linux distribution from an available distribution and include the use of regular file systems available on Linux like XFS and EXT3. For our protocols we use implementations like Jetty, Apache Slide, kernel NFS, SAMBA project, etc. So I applaud your company’s release of a solution which is based upon open standards as well, welcome to the club we’ve been here for a number of years already.

So I think that a I answered your question above, and I do want to point out that Hitachi is and will continue to exist. There are many reasons for this which I won’t get into, but I will point out that while the US allowed Lehman to fail, note Nomura a Japanese company bought a lot of the Lehman assets, it did not allow GM to fail. Further the comparisons about sgi and SUN to prove your point are not well thought out in my opinion. sgi has long been known to be on the skids since well before the current down economic cycle, and Sun/Oracle well they are a strong partner of Hitachi and have been most recently in the stage of reinventing themselves with the originating change point being the dotcom crash. So the sky is not falling and I think that my original post still stands.

Thanks for the reply Michael, I am sure you customers will be re-assured by your personal guarantee that Hitachi will be around forever.

You’ll have to excuse my “ignorance” (coming from a small principality the wrong side of the pond etc), however, I am sure that if you re-read my comment a few times or even the blog post you will probably find that I did not give specific examples of any product line from any vendor. Many organisations out there are facing a wide array of proprietary solutions and platforms out there that lock them to a vendor? If HCAP does run on off the shelf tin, standard OS and FS then more power to your elbow.

In the video space there are 1000’s of organisations creating 10s to 100s of terabytes of content every year. They too need to guarantee content authenticity and longevity. They too need to ensure content is immutable and protected from deletion. Can they afford to pay for the Hitachi legacy to be maintained? No.

As you bring up HCAP let’s see just how affordable HCAP is using the pricing lists over on StorageMojo (we have to use list pricing as the comparison as it is the only information we have):

A 10TB HCAP system is going to cost you $150k to $200k once all the “extras” are taken into account. $18k just to search 2.4TB of content? (Odd since Archivas was a $15/GB searchable object based archive solution).

http://storagemojo.com/storagemojos-pricing-guide/hitachi-storage-price-list/)

Even if the price list is a touch out of date I am sure it is in the ball park.

There is clearly space in the storage market for both models. What will be interesting is which model the market will lean towards in the coming years. My hunch is that the majority will choose from the swell of commodity based storage solutions (like BackBlaze, MatrixStore etc) that give customers what they need at a cost that they can afford.

Nick I started replying and it got too long please check out here for a response to your comment.

[...] about Om Malik as of October 3, 2009 Chapter 3 – Cheap Cloud Storage, but What about TCO? – blogs.hds.com 09/29/2009 So I’ve read and poked around a bit on BackBlaze’s cheap [...]

Post a Comment