Follow-up to Previous Post
by Michael Hay on Oct 2, 2009
Nick commented on my comment back to him on my previous post and my second comment got too long so I decided to dedicate an entire post to the response.
Nick this was one of the points from your second comment to me:
“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.“
I do take issue with this. While I agree you did not reference a specific vendor in the original comment to my post, you did in your blog post you referenced from it, here is a direct quote:
“What do you do with an EMC Centera, a HP RISS, or an Isilon system if you decide that you don’t want to use their software anymore, or if they discontinue that product line? You’ll end up moving data off the system whilst ending up with rather a lot of junk metal.“
I think that the above quote shows that indeed you were referencing both vendors and products in your materials.
As to your points on pricing, the Mojo’s Hitachi list is 15 months out of date and actually predates the HCAP-300 (RAIN based) model expanding to 45TB, Mojo says it is from July, 2008. Further the WMS100 was removed from all new HCAP packages, largely with HCAP-300 taking over the canned appliance space. As for our SAIN configurations they now use the AMS2000 series or the USP-V(M) series if customers already have these footprints on their floor or like the added assurances that comes with a SAN deployment. For instance if they want to have a completely decked out file strategy they could have HNAS, HCAP and HDDS all on one AMS2000 or USP-V with HDP running to improve capacity utilization and performance, they can and there are added benefits like storage tiering. (Check out David Merrill’s blog and you should consider potentially contacting him for discussions on Storage Economics including storage tiering. While I cannot commit him I’m sure he would be more than willing to discuss the ROI model that you are working on. http://blogs.hds.com/david/)
The other thing that I want to get back to is that for every $1 of capital expense that is incurred there are $7-$8 in O&M costs. Further IDC — note the above information is from IDC’s latest server report — predicts that over the next several years capital expenditures will either decline or remain flat with energy consumption and operational costs like patch management, etc. costing the most. This was the original intent of my post and I think that barring the mention of Hitachi Nick, I would ask you to state if you agree or disagree with my point: namely that TCO is a major factor when considering costs of any deployment? For example if it costs you $3/GB to buy the system, but with a fully burdened cost the actual number is $20/GB due to needing labor to nurse along a the system was that a wise investment?
While I cannot provide the customer name, I will say that for their deployment of a parallel filesystem, Lustre, 18 administrators are required just to keep the filesystem operating. Now I’ll be the first to admit that I don’t know the salaries of these employees, but I will just put sane round numbers out there to prove a point, if we blend the skills and assume that the salary average per employee is $100k/year (some are $120k and some are $80k) then that is an extra $1.8M/year in employee costs. If one of their vendors could offer a functionally equivalent COTS solution and was able to improve the usability and manageability of the system then they should need less personnel and potentially less skilled personnel the manage the system. These are exactly the kinds of things that vendors do: build coherent packages, improve usability, reducing management costs, improve densities, improve scalability, etc. all of which reduce TCO. (Where I think that Nick and I are in disagreement is on whether or not either company can survive the current economic downturn. Fortunately, both companies practice an open approach that puts the control of the data in the hands of our customers. After all as Nick stated that this is critical for any vendor to do and both MatrixStore and HCAP take this approach!)
Nick the last point that I want to talk about is the work ignorant. Here it is taken from the Oxford English Dictionary:
• adjective 1 lacking knowledge or awareness in general. 2 (often ignorant of) uninformed about or unaware of a specific subject or fact. 3 informal rude; discourteous.
— DERIVATIVES ignorantly adverb.
— ORIGIN from Latin ignorare ‘not know’
Here I take both the Latin point of origination and the second (2) definition above. So I literally meant that you did not know about Hitachi’s HCAP. In fact on a daily basis I’m ignorant of something and happy about that point. I live in Japan and I’m learning the Japanese language so in a very real sense I’m working to remove my ignorance of the language on a daily basis. So this was not meant as an insult in the least, it was an honest statement that you did not know about our offerings.