My original response to Chuck

Again, I apologize to my regular readers, but dang, stating factually incorrect things, just pisses me off to no end. Chuck won't let me respond in his blog, so I am forced to waste bandwidth here... This was my original response to Chuck which I crafted last night. You'll notice I was a lot calmer when I wrote this one.

NetApp's guarantee, guaranteed a Chuck Hollis rebuttal.

And although I would love to put my rebuttal inside of his blog, Chuck also made it clear, I was not welcome:

I'd be interested in any and all opinions -- that is, from people other than NetApp employees.

[Recently updated: Chuck now let's NetApp Employees post to his blog, sadly too late for this blog ... .]

Thankfully, I have my blog, and for what it's worth, Chuck and any EMC employee is always welcome to comment on my blog.

Chuck had a lot of problems with our guarantee, but I couldn't get past the first problem he had to read the rest.

You must use RAID 6 as compared to  "traditional" RAID 10.  Now, we can argue about the pros and cons of each, but most storage people would not consider them directly equivalent in terms of protection or performance.  Never mind that you can get RAID 6 on all sorts of storage arrays today.

So here's what he said:

1. You must use A  and not B
2. A may be better than B or B may be better than A
3. Collective wisdom says B is better than A
4. A is available everywhere, so what's the point of saying you must use A.

To the best of my understanding, here's what he was trying to say:

NetApp's say you have to use RAID 6, which performs worse than RAID 10 (despite benchmarks results that you should ignore) is less resilient than RAID 10 (despite proofs you should ignore) and therefore is inappropriate for most workloads.

And if by the way if this whole guarantee hinges on that RAID 6 thing we can do that too. (Well except if you need performance. Our RAID 6 performance is 60% worse than our mirroring performance, but you don't want RAID 6, you want RAID 5 with active sparing, except if you want big disk drives and resiliency and you don't want that you want EFD's).

Well, okay....

Comments

[Mike Shea]

After about 6 years at EMC, I understand a few things about them - and I'd like to help.

If EMC does not have something, it is characterized as "bad". Only when they finally can deliver a function (like thin provisioning) is it now 'good'.

So, why are they calling a guarantee 'bad'? Because they cannot provide one. Same reason that they are calling dedupe on primary 'bad'.

Competing with EMC is a lot like studying abnormal psychology. You may not really understand the afflicted one, but at least you can turn it into an academic exercise.

But enough of all this - as T. Boone Pickens said - "Stop chasing Rabbits if you are hunting Elephants".

[Chuck Hollis]

Oh, BTW, feel free to post a rebuttal over at my blog.

I guess I mis-spoke. I knew you guys would be weighing in.

I just wanted to invite others to express their opinion, since y'all have this habit of gang-tackling blog posts you don't like.

I'll go change it to be more clear. C'mon over, the water's fine!

Gee, you're so *sensitive* these days ;-)

-- Chuck