« Towards Thoughtful IT | Main | EMC's New DW/BI Competency Center »

December 09, 2008



"Some work out well, some don't -- but there seems to be a grudging consensus that the Dell/EMC relationship has done a good job of re-shaping the dynamics of not only both companies, but the industry itself."

i would say the grudging consensus is that EMCs share of Dell revenue is decreasing and the growth of that channel is decreasing as well. i would hardly say that the conesnsus is that this acquisitino or grab for some of those fat storage product revenues by Dell is good in any way for EMC, indeed its what sparked a share price decline from 25$ to 15$ before the financial markets imploded. now i think that the dell acquisition of equalogic will be a total disaster and they will fudge it up royaly and long term Dell will suck a big one in the storage industry, but 100% guaranteed the conesnsu is that it was not good for EMC and the numbers show your relationship with them has changed since Dell bought that company.

Tony Asaro


Question - the NX4 directly supports NAS and iSCSI but the FC is direct to the AX4? Is that correct? Do you configure NX4 with iSCSI and NAS with one management software and the native FC with another? How many steps in the configuration process? How about troubleshooting and continued maintenance - is the management software the same regardless of protocol?

Also - I believe that the NX4 iSCSI is implemented in the same way as NetApp - block on a file system - therefore the NX4 is not doing "native" iSCSI using your way of thinking. Yes? Therefore - using your logic - EqualLogic has a better implementation of iSCSI than the NX4.

Have you thought about that the FAS with FCoE will unify the network with all storage protocols resulting in lower cost and easier management?

Additionally, do you believe that emulating FC (as you call it) is in the same ballpark as emulating a mainframe?

I am confused about the market share numbers - according to Gartner NetApp leads substantially in pure NAS and unified storage.

Chuck Hollis

Hi Tony

I'll assume for the moment that these are sincere questions, and you're not just throwing rocks again.

The FC is native, and is not mediated by the NAS head. The software tool that provisions all three environments is relatively integrated.

In some special cases, people will want to talk directly to the underlying array, which they can do if they choose. You're welcome to a demo at some point, if you're sincerely interested.

You are right that the NX4's iSCSI is implemented as an emulated layer on top of an underlying file system, much as is the case with NetApp.

But you lose the thread there -- most FC care about predictable performance behavior, and emulation can often get in the way of that, as you lose the underlying geometry and device characteristics.

Conversely, iSCSI is rarely about nosebleed performance, and is more about low cost and ease of configuration and management. It also offers a very decent level of performance in more use cases than most people realize.

And, of course, if someone doesn't want that flavor of iSCSI, all EMC platforms support a native iSCSI as well that doesn't require a NAS head.

That's the beauty of having a very robust portfolio, we don't have to get into arcane arguments about which is best -- people have choices.

Regarding FCoE, I've done about 4 or 5 posts over a long time. I think the first one was over 18 months ago, and I've been updating my readers continually with progress.

You're right -- not only is there converged hardware, but the potential for converged management as well.

You might want to read the post on "Who Will Manage The Converged Fabric?" for some of the interesting wrinkles we expect to see on this. I'd be glad to update you at some point as to what's happening in this part of the industry.

I used the mainframe example simply as an analogy. All analogies break down at some point.

That's why we call them analogies.

We tend to use IDC numbers internally here at EMC as we understand their methodology and they tend to accurately reflect what we see going on in the marketplace.

They give us a nice, sophisticated breakdown that enables us to do all sorts of useful pivots and analysis. Great data sets which I highly recommend if you take this stuff seriously, which we do.

I can't speak to Gartner's numbers, or any differences between their perspective and IDC's numbers. If you like, I can connect you with EMC's research analysis team to share their perspectives regarding the differences between the two.

-- Chuck

Tony Asaro

Chuck -

You mentioned two things - predictable performance as a requirement of FC. But then you mention iSCSI isn't required for "nosebleed" performance. I don't see those two things as the same.

As we both know - performance is impacted by a number of things - the types of I/O workloads, processing power, caching algorithms, the speed and amount of memory, types of disk drives, how data is stripped, bandwidth, etc. I don't see any evidence that "native" FC will outperform "emulated" FC based on all the variables impacting performance.

In terms of predictability - now that is interesting to me. I have not heard of any cases where NetApp FC didn't provide this. I am seriously asking this - and will ask NetApp as well - doesn't the FAS file system undestand the underlying geometry and device characteristics?

Listen - on the throwing rocks thing - you believe what you will. I find it ironic that you of all people would accuse anyone of throwing rocks - you really should take a long look in the mirror.

- Tony

Let me ask you - do you really believe that NetApp's implementation of FC is not valid? If the answer is yes, then how do you rationalize the fact that they have a large number of implementations?

Chuck Hollis

Tony, sorry, but it's pretty apparent to me that your new gig involves projecting an extremely hostile stance towards EMC -- and me personally -- which is most likely a key part of how you're marketing yourself to the other vendors.

Regarding your observations, I have to say -- with all due respect -- you're not in the trenches.

My guess is that you've never built this stuff, installed this stuff, supported this stuff, or run this stuff in a large environment.

If that's true, that would make you an armchair analyst, my friend.

I could shower you with all the nice docs that are out there, but why bother? We have a nice business at EMC ripping and replacing failed NetApp FC environments that have been seriously oversold by either NetApp or one of their partners.

It's not the ideal way I'd prefer EMC to make money, to be honest.

You might want to ask someone in a classical large data center if they'd consider putting their most demanding apps on NetApp FC, and watch their response.

Regarding disk geometry, NetApp (as do all emulated environments) attempts to emulate physical disks using a file system.

One could argue pedantically the pros and cons of such an approach, but the bottom line is that behaves very differently than native FC, especially when the application is demanding, the array gets full, you're depending on sequential streaming, or some other condition.

Again, if you're being sincere, I'd be more than happy to hook you up with more details, and some very knowledgeable people.

Best of luck with the new gig, Tony.

-- Chuck

Martin G

"You might want to ask someone in a classical large data center if they'd consider putting their most demanding apps on NetApp FC, and watch their response."

It really depends on what you call most demanding doesn't it? I'd be interested to know the definition to be honest because I know of some very large organisation who have very large Exchange environments running on NetApp FC and appear to be very happy with it.

I suspect 90-95% of my workloads would run happily on NetApp but I have a large investment in EMC and it doesn't make a huge amount of sense for me to change presently and anyway, I quite like DMX.

And yes there are 5-10% of my workloads that due to their I/O profile that I probably wouldn't put on NetApp but as long as an array performs in an appropriate manner for the application I'm running on it; I couldn't care less about whether it is emulating or native.

What does 'relatively integrated' mean for the management tools? As far as I know you don't have a unified management interface for the Celerra which allows me to manage FC,iSCSI and NAS using the same tools/CLIs. When you have, you can at least pretend to have a Unified Storage Platform; ramming two products into a box and claiming that they are a unified storage platform is a bit rich in my opinion.

Yes, it is great that you can take FC out of the back-end Clariion in the Celerra; potentially very useful for some shops and it doesn't stop either of them being good products but please they run two internal operating systems in DART and Flare and are managed using different tools. And if you have a new wonderful integrated tool, please, why aren't you telling me about it?

I remember various people ramming PC-Cards into their workstations/proprietary workstations/servers but it didn't mean that Windows was integrated in any meaningful manner with OS/400, Solaris etc.

VMWare Fusion is the closest I've seen to an seamless integration into another operating systems; especially running in Unity mode but you still have to use the Windows management tools at times. I'd argue that from storage point of view, you aren't even this close with the Celerra/Clariion integration.

Chuck Hollis

Hi Martin -- always a good conversation, so lets get started, shall we?

I'm sure that -- somewhere -- there are people running Exchange, SAP, etc. on NetApp FC. And they may be relatively happy.

My point is quite different. We are doing more than our fair share of "second surgeries" these days on oversold NetApp FC implementations these days. That's not a pleasant experience for those involved. I think this is a bad thing.

My comments regarding "relatively integrated" are directed to the config wizards used on the lower-end Celerra boxen. I'm guessing you're in a much larger environment, and are using different tools altogether. If you'd like to take a close look at an NX4, that can be arranged, I'm sure.

Regarding your thoughts on what is unified storage, and what is not, I consider that a relatively academic debate.

I tend to take a more pragmatic view. Just like there are different ways to do snaps, remote replication, tiering, etc. I don't think we live in a world where there is one and only one interpretation of a unified storage concept.

I would expect different vendors to come up with different approaches to the "all in one box" concept.

Unless you're a member of the Church of Netapp, that is.

I too like VMware's Fusion. I lost you a bit when you started contrasting it a bit to Celerra and CLARiiON. If you just consider the iSCSI emulation on Celerra, though, it's a pretty fair comparison.

The real debate is --- when you need FC and what it does well, are you better off with an emulated version, or a native implementation?

And, remember, if you don't need what FC does well, there's always iSCSI, which I believe is generally underappreciated in enterprise shops.

Based on your comments, I think your purchase decisions tend to reflect that thinking.

And, of course, everyone's situation is different.


Alex McDonald

@Tony "I am seriously asking this - and will ask NetApp as well - doesn't the FAS file system undestand the underlying geometry and device characteristics?"

Yes. I know of no storage system that talks to disks directly that doesn't.

What NetApp does is virtualise the underlying storage to present a virtualised LUN in the SAN case. EMC's Navisphere Virtual LUN technology also presents something that doesn't reflect the underlying geometry and device characteristics either.

Perhaps the real vs emulated FC analogy broke down a lot earlier in the discussion than some would care to admit.


This post and your comments have more spin than a top:

You remind me of a very well known -1990s- band called Spin Doctors: http://en.wikipedia.org/wiki/Spin_Doctors.

I think Tony deserves a bit more credit for his "new gig" -as you like to call it, condescending if you ask me (and that's not meant as a personal attack at him because you never do that, right... http://blogs.netapp.com/exposed/2008/09/blogruptcy.html#comment-129524912 ) - because he poses a few very valid questions, whether you like it or not. I tend to agree with him on this one.

Chuck Hollis

Hi Sjon -- sorry if I'm expressing my personal opinions here. If you call that spin, so be it. That's what blogging is all about, right?

Tony's first foray in his new gig was a paid-for attack against EMC on HDS's blog site. The questions weren't reasonable, it was a competitive hatchet job.

I responded vigorously.

I detest the idea of pay-for-blogging by vendors, especially by those who try to market themselves as "independent analysts". I'm all about transparency, and it was pretty clear that Tony was trying to hide the true nature of his new assignment.

By comparison, if you take a look at my blog page, it's pretty damn clear who I work for and where I'm coming from.

There's a difference in my mind between valid questions, and someone trying to work a competitive agenda. I can tell the difference.

Can you?

BTW -- really liked the Spin Doctors in their day.

Chuck Hollis

Hi Alex

Tony is confused again here -- he misinterpreted my statement, and started asking irrelevant questions of EMC and NetApp on Twitter.

Of course, every storage OS understands the nature of the drives they talk to. There are gradations of how well they do this, but that's not the point here.

My point is that -- at the application level -- emulated FC drives behave differently than real ones. And, in many cases, it can matter.

I don't really have the time or the interest to go sort Tony out on this one. He's still trying to explain FCoE to me :-)

-- Chuck

Alex McDonald

"My point is that -- at the application level -- emulated FC drives behave differently than real ones. And, in many cases, it can matter."

Drives? Are we talking protocol out the front end or disk connectivity at the back? A LUN isn't a drive; it's a *logical* unit, not physical. Please enlighten me as to what you mean; your use of terminology is causing a great deal of confusion.

Chuck Hollis

Sorry, Alex, you're right.

I'm mixing specific protocols and use cases. To an engineer, FC is a protocol. To a marketing guy, FC is a specific use case -- at least, compared to iSCSI, NAS, infiniband, etc.

Another example from our distant past? 3270 is a protocol. It's also a very specific use case, if you think about it.

I have no basis to comment on NetApp's specific implementation of the FC protocol. Unless I talk to one of our FC engineers, I just wouldn't know. Or know if any deficiencies or advantages really mattered.

I do have a basis to comment on viability of NetApp's FC implementation in many FC use cases. To be fair, there are many shops that have a bit of FC in their environment for whatever reason -- test and dev, legacy application or server, etc. No complaints with any of this.

My complaint has to do with environments where people tend to hammer their arrays -- a very common FC use case, which had nothing to do with protocol specifics.

I hope this clarification helps?

Tony Asaro

Just for the record - I am not an independent analyst and have not positioned myself as such in my "new gig". I am a consultant that works for clients that I select. I have been candid about that from the beginning.

Chuck Hollis

Thanks for the clarification, Tony. I appreciate it.

charlie lavacchia

Hi Chuck,

Would you kindly share with your readership those elements in your background that would indicate that you yourself pass your own "Armchair Analysts" test?

- Have you ever built this stuff? From my perspective, that impliess having been a storage hardware and/or software developer.

- Have you ever installed this stuff? From my perspective, this implies having been a hands-on SE i.e. not in mgt but with real hands-on install experience.

- Have you ever supported this stuff? From my perspective, this implies having had a hands-on break/fix and/or a hands-on Professional Services role (hands-on again being the real practioner vs. armchair differentiator).

- Have you ever run this stuff in a large environment? From my perspective, this implies having been a storage admin in a reasonably large environment.

I ask as I intend to use the same conditions you call out as an "Armchair Bloggers" test.


Charlie Lavacchia

Chuck Hollis

Hi Charlie -- I think that's a fair question.

I thought you might know this -- coming from EMC and all -- but I suppose I could update you.

I'd point to the following:

* Silicon Valley geek from an early age -- first programming job was in high school.

* got CS and Econ degrees from UC Santa Cruz

* UNIX kernel and I/O stack work in the SF Bay Area as a software developer early in my career -- my specialty at the time was disk device drivers and scheduling algorithms.

* technical marketing of large UNIX servers in dbms environments later on, including an extensive stint doing benchmarking.

* a short stint with an unnamed database software vendor doing technical marketing.

* helping EMC get into the UNIX storage market in the mid 1990s (included application profiling, use cases and benchmarking)

* direct hands-on engagement with a number of EMC's larger customers while we were introducing Symmetrix for open systems

* ran a technical presales and consulting org in EMEA for EMC at the end of the 1990s

* was brought back to EMC to work with product teams to respecify EMC's storage platform offering, resulting in (among other things) the DMX, CX, Celerra and Centera product lines. Also convinced EMC to get into the disk library business, the result was the current EDL.

* and a bunch of other stuff that really isn't public knowledge, and I don't want to share :-)

No one person is an expert, but I'd offer that I've done enough hands-on (and worked closely enough with hundreds of people that do all hands-on) that I think my background is reasonable enough for the topics I cover -- over 18 years with storage and related topics.

Hope that helps!

BTW, where are you these days?

-- Chuck

charlie lavacchia

Hi Chuck,

Thanks for your response - that is quite an impressive resume!

During my 7 year tenure between 1994 and 2001 our paths didn't truly intersect until 2000. As we both know, as EMC grew to 20K strong, it (like many tech companies) had its share of poseurs during that period. I state this from personal experice - I worked for a few. It is gratifying in retrospect to know that you weren't one of them.

(Full disclosure to your readers - I worked for Chuck in 2000 / 2001 and it was Chuck who laid me off in May 2001.)

As I wrote under my name at the bottom of my comment, I am with NetApp. I am in my 6th year helping make many former EMC FC SAN customers quite happy as they transition their SANs to NetApp's uniquely efficient and truly "Unified" storage platforms.

Happy Holidays!

Charlie Lavacchia

Chuck Hollis

Good to hear from you Charlie, glad you've found a gig that works for you.

And, just to be clear, EMC does a healthy business replacing NetApp gear that's become inefficent, functionally lacking, insufficiently robust or crumbles under a demanding workload :-)

Not to mention virtualization, security, content management, cloud -- or any of the other businesses we're in these days.

I wonder who's winning the race?

Have a great holiday!

-- Chuck

Used/Refurbished laptops

I must say I have been suprised that (until now) neither Dell nor Apple have had direct communication with the owners of laptops subject to the battery re-call.(assuming my family is a good litmus test - OK we live in Australia so maybe we're not a good test!!)

EMC Legato EmailExtender Rube Goldberg Edition

What the frsk do batteries have to do with a good old fashioned storage smackdown?

Now I could understand it if a misguided chucklis-tar (love the flames in the background. At first I thought they were black, but hey, I'm colour blind. My daughter told me they were red. Got it.) salvo triggered your fight/flight response; that's what smackdowns are all about. Outright spam now... well, even Chuck doesn't deserve that.

refurbished computers

I guess the best time to sell storage is right alongside the server decision.

Used Servers

Just wanted to chime in as a customer.

Just picked up slightly used Dell NX4 because I really love our Dell AX4-5i.. the only problem is drive upgrades are quite expensive.

As a tiny niche public cloud startup we wanted to invest in enterprise level hardware that wont crumble just when we start gaining momentum. Our cloud was initially designed to run on used commodity hardware... and it does in terms of hypervizors. Hypervizor failure will not effect the cloud in the least due to automatic transfer to another hypervizor.

With SAN all our experiments with open source and paid san software and commodity server hardware did not pass any stress tests or benchmarks so we decided to go with AX4 due to low pricing and excellent reviews.

AX4 has been working great and we are proud to mention to our customers that we are running on EMC as it gives them a sense of security.

The comments to this entry are closed.

Chuck Hollis

  • Chuck Hollis
    SVP, Oracle Converged Infrastructure Systems

    Chuck now works for Oracle, and is now deeply embroiled in IT infrastructure.

    Previously, he was with VMware for 2 years, and EMC for 18 years before that, most of them great.

    He enjoys speaking to customer and industry audiences about a variety of technology topics, and -- of course -- enjoys blogging.

    Chuck lives in Vero Beach, FL with his wife and four dogs when he's not traveling. In his spare time, Chuck is working on his second career as an aging rock musician.

    Warning: do not ever buy him a drink when there is a piano nearby.

    Note: these are my personal views, and aren't reviewed or approved by my employer.
Enter your Email:
Preview | Powered by FeedBlitz

General Housekeeping

  • Frequency of Updates
    I try and write something new 1-2 times per week; less if I'm travelling, more if I'm in the office. Hopefully you'll find the frequency about right!
  • Comments and Feedback
    All courteous comments welcome. TypePad occasionally puts comments into the spam folder, but I'll fish them out. Thanks!