A Quick Update From NetApp

As I was scanning the morning emails, I found a link to this gem, basically pre-announcing NetApp's announcement for tomorrow.  The biggest surprise was that there really wasn't much of a surprise -- just about everything they announced was already public knowledge.

However, working as I do for EMC, I'm somewhat obligated to offer up some sort of commentary when one of our storage competitors makes an announcement.

Ready, Fire, Aim

I guess the gun went off about a day early for the team at NetApp.  Seems that one of their international sites posted launch details early, which caused one of the journalists to break embargo, and -- once that happened -- everything was out in the open.

Doesn't really matter in the big scheme of things, but I bet there are some NetApp people who are besides themselves with frustration.

Their big announcement theme is around "making the leap to a shared IT infrastructure", which -- to me -- is quite puzzling since IT groups have been consolidating and sharing IT infrastructure for quite some time -- like for more than a few decades.  Yes, widespread convergence and virtualization are driving fresh discussions in the industry, but the choice of Big Topic strikes me as well, conservative.

Fresh Tin

NetApp customers have been waiting patiently for some fresh tin: the existing storage controllers were getting very long in the tooth, especially with the rapid pace of processor evolution.  Put differently, it was getting ridiculously easy to win most any performance benchmark against NetApp.

I'll reserve final judgement until we get some of the new models in house and put them through their paces, but -- for now -- I'll give them the benefit of the doubt and offer up that they've probably returned to relative competitiveness in terms of storage controller performance.

Finally, SSDs

I think most of us got fairly tired of the NetApp party line: NetApp didn't need SSDs like other vendors because they were "special".  Those of us who were industry insiders knew that -- sooner or later -- they'd have to offer an SSD option to their customers.

I don't have the details, but from the initial materials, it looks like a pure-SDD shelf -- no intermixing of media, and nothing like FAST with automatic tiering.  I suppose it's better than their previous approach of trying to put their V series in front of a TMS RAM disk, but still ...

Some New Big Boxes

You've probably noticed that all the storage vendors are pushing their drive counts to unheard-of levels.  NetApp's new models are no exception to this trend -- they now are claiming up to 1440 drives in their big boy arrays.

While I might be tempted to give these new models the benefit of the doubt for low-IO-rate archival use cases, I think it's going to take a while before most customers would be confident putting that much data on a brand-new array with brand-new software.  We're somewhat fortunate at EMC that we've been offering 1000+ drive arrays for quite some time, so I can offer up some intelligent guesses as to some of the teething challenges they'll likely face.

One test I'll ask our engineers to run: how long does it take to do a traditional restore on these beasts?  

Data restores are brutally write intensive for sustained periods, completely overwhelming any caching or pointer tricks the array might be using.  It's one thing when a terabyte restore takes far longer than it should -- it's another thing entirely when we're talking a petabyte-class restore slogging along at a small fraction of more typical speeds. 

Perversely, you really care about restore performance when you need it ...

And, yes, I've completely lost the plot as to the different modes of their ONTAP storage operating system, and which one does the scale up and which one does the scale out, and the ever-changing list of restrictions and compatibility issues between the two.  

I'm sure someone will set me straight (again!) before too long.

A FlexPod?

As expected, NetApp is cobbling something together to attempt to compete with the VCE Vblock.  I've already had my say about the current raft of Vblock clones here, so I'll spare you.

All I really have to add is my personal prediction: you likely won't hear them talking much about FlexPods next year at this time -- it'll probably be gone and forgotten due to lack of customer interest.

The Bottom Line

Nothing too surprising or game-changing here, from where I sit.  

Some predictable updates to aging tin, a re-announcement of their operating system, a bit of catch-up in the areas of SSD and non-disruptive data migrations, and an attempt to respond to the uber-successful Vblock.

Personally, I was hoping for something a bit more daring or unusual from NetApp to push the boundaries a bit -- perhaps announcing they might be getting into an adjacent business like backup, or security, or even data warehousing.  Or, even staying with storage, perhaps announcing something that had the potential to be game changing.

But no -- seems like they're playing it safe for the time being.  

Dare I say it -- has NetApp become boring?

Comments

[Jeremy Welling]

Disclaimer: I work for NetApp

I'm too lazy to respond to your whole post, but I'll clarify one thing for you Chuck, and ask a few questions.

NetApp previously had Ontap 7 and Ontap GX. Ontap 7 was our legacy codeline and GX was our scale-out clustered codeline. Now we have Ontap 8 - 7mode and Cmode (cluster mode), corresponding to Ontap 7 and GX. This is our 1st step to converging the two. It's that simple.

Now that I've clarified that, riddle me this Batman: How in the heck do you keep up with the Dart, Flare, Rainfinity, Centera, and Data Domain firmware? EMC ships different firmware versions for every piece of hardware they make, and it takes all of them to try to match the features of Data Ontap.

I have customers that needed Excel spreadsheets to try to keep track of your code revisions. You throw flack at NetApp for a single firmware release with two modes while your kit is cobbled together from so many completely separate products I can't even begin to keep track.

I know, now you're adding a "Unified" front end on it to mimic what NetApp has done all along. As I said to the local EMC SE the other day, lipstick on a pig...

Cheers,

-JW

[Chuck Hollis]

Hi Jeremy

The products you name do not have equivalents in the NetApp portfolio -- they're specialized storage platforms.

For example, there's no functional equivalent of either Centera or Atmos in the NetApp portfolio. No functional equivalent of RecoverPoint that does CDP or CRR. No dedupe backup target device like DataDomain. Or VMAX or VPLEX for that matter.

This list of "no equivalent from NetApp" goes on for quite a while.

NetApp spends a lot of time and effort trying to convince people that either (a) these functions aren't needed, or (b) whatever you might have is good enough.

Also, keep in mind that we're entirely different companies: EMC is many times your size, and may times your market share.

People expect EMC to have good answers to difficult storage requirements -- they don't expect that from single-product companies like NetApp.

We generally train our folks to lead with our unified storage product as a single platform that does as much as possible for a customer without introducing unneeded complexity.

For those customers who can get by with a single storage platform, that's great.

However, if they've got a specialized challenge: archival, replication, content mgmt, etc. -- we'll we've got specialized platforms for those. Your company doesn't.

My question would be the opposite: what does NetApp do for customers that really need these functions?

I think the answer exposes a key difference between a company that sells a single product, and a company that tries to solve complex customer problems.

-- Chuck

[Jeremy Welling]

Chuck,

NetApp solves many complex customer problems, including replication, archival, deduplication on primary SAN & NAS, VMWare on SATA, etc. In fact, we have a wide array of flexible solutions which render much of your purpose built products obsolete.

Take for example the recent developments with our Syncsort partnership. That solution trounces your Avamar-DataDomain combo on simplicity, cost, and features. The first time I saw the demo of BEX doing a P2V conversion on the fly from the backup data on NetApp storage I was stunned. Bare metal server recovery of a physical to a VM in < 5 minutes? Easy failback later? Heck yeah!

Chuck, to point at Data Ontap, NetApp's core enabler technology and state we only have one product is misleading. It's tantamount to pointing at Microsoft and stating they only have one product because their core technology is Windows. /facepalm

-JW

[Paul]

seriously Chuck...bless ya... that'll be why you work in marketing...poitless post with nothing to say to be quite honest..lets face it the reason you have some many boxes is they are cobbled together by aquisition after aquisition, no commonality of platform or function...want a NetApp solution to backup...well just turn on the feature...want a dedupe box...well heck it already is...deduping production data...quick scan for EMC production dedupe...no...must of missed it!..

heck...already wasted to much time bothering with this...have fun Chuck...keep up the work...but really you must be able to right better than this lazy stuff in future...

[Chuck Hollis]

Hi Paul

Honestly, you NetApp folks need to get out in the big, real world more often. There is a universe of use cases and workloads where the NetApp products in their current form just don't meet customer requirements.

No amount of happy-talk is going to change that fact.

BTW, EMC's unified arrays have been deduping (compressing) production data for quite some time.

See? You *do* need to get out more often.

-- Chuck

[Chuck Hollis]

Hi Jeremy

You sound so happy in your self-contained world.

Don't want to disturb you with inconvenient truths (like 5 minutes for a P2V bare metal restore is *way* too long for many use cases), so I'll let it lie.

Best of luck.

-- Chuck

[Jeremy Welling]

Chuck,

At least gift us with some sort of technical response if you're going to discount the solution.

What EMC (or partner) product(s) would you suggest to provide better functionality in this category? On the fly heterogeneous physical to virtual conversion, firing up live off backup storage, automatically registering with VMWare?

I'm not talking about a server HA solution, as that's a different category. I'm talking about a cheap and flexible way to utilize unified storage to add resiliency to an environment and provide a single pane of glass for replication infrastructure and DR.

As an aside, I will confirm, I'm quite happy with a 4 petabyte, 30+ site, 7 continent, single command DR failover, WAN replicated, no-tape, no vtl, VMWare, deduped, thin-provisioned, snapped, cloned, unified, flexible, cloud-enabled, "small" world. I'm professional services, so I'm well aware of what life is like in the real world Chuck. You may want to check it out yourself sometime...

-Jeremy

[Chuck Hollis]

Why is it that most NetApp people are so proud of their boxes, and never, ever mention what the customer is trying to get done?

Maybe we should call this "The Little Filer That Could".

Jeremy's example above (if it's real, that is) begs for a preamble around the customer's real-world use case, and not a bragging list of features.

-- Chuck

[Jeremy Welling]

#1 Unified, easy, and efficient backups with site replication and DR (I would think that should be obvious).

#2 The Ontap 8 cluster I deal with clocks 1.7 gigabytes/second sustained random read/write. It's a "little" busy.

#3 It's real, go watch the videos on youtube. Since you apparently lack some imagination, here are some scenarios.

Say your primary data center is going down for a power outage and you want to fire up some critical physical servers as VMs at your other data center? Solved. Your sprinkler system turns on and half your data center is shut down but your VM farm is up? Solved. You want to eliminate VTL or tape from your remote offices while improving resiliency? Solved. Block-level heterogeneous incrementals, dedupe, compression, unified replication, and DR. Solved. Five minutes max to fire up any physical server in your data center or remote site as a VM? Wicked cool. How much more real-world do you want Chuck?

Best of luck,

-Jeremy

[Jeremy Welling]

I've been reminded I left out one of the best scenarios - P2V migration.

Say you've got a hundred 8 core physical nodes which are coming off lease. It's been decided these are now moving into your VMWare cloud. In preparation, the customer adds the appropriate VMWare hardware. BEX then handles the conversions, complete with automatic MBR alignment. A year down the road, it's time for the 12 core servers, etc. etc.

There are plenty of other ways to accomplish P2V scenarios. Are they that easy though? I mean, an integrated feature of using the right software and storage? No sir.

[Kees Henniphof]

Well mr. Hollis, if NetApp's portfolio is as marginal as you seem to think, then please explain why you had little EMC trucks with big fat advertising driving around our European headquarters all day long on Nov 4th, when we hosted a press conference in support of our Future Ready IT launch - the introduction of our new platforms and management software. Apparently, somebody got nervous in your European marketing team? Pls find details here: http://wp.me/pNUTu-3h

[Barnacle Bill]

Chuck,
I must say, you're quite mistaken: Netapp is not a one-WAFL diner. Netapp has a very diverse product line covering many different business requirements. For instance, there's Netcache, an Internet content solution. Alacritcus, a VTL product. Data Domain, a better VTL product (oh, wait, scratch that). Spinnaker, a scale-out solution. Decru, an encryption solution. Topio, a CDP solution that is absolutely *not* a Recoverpoint clone. And finally Bycast, a scale-out, object oriented storage system that is absolutely *not* an Atmos clone. Hmmm...except for Bycast, I guess all those companies are dead now. Sometimes it seems like Netapp tries and tries to diversify by acquisition, somehow doesn't find out through due diligence that product X doesn't run Ontap, kills the product/company once everybody realizes it can't be shoehorned into a 17-year-old kernel, and finally ships a press release that amounts to, "We don't need no stinkin' encryption/scale out/VTL!" But Chuck, at least give them credit, 'cause they keep on spending the money and they keep on trying.

[Chuck Hollis]

Hi Kees

We do various competitive marketing things with a wide array of EMC's competition. That would include NetApp among other vendors.

-- Chuck

[Mark Edwards]

Chuck,

Like the Nancy Pelosi, you are helping your competition grow and thrive.

Mark

[Neil Fletcher]

Chuck

I'm not in the market for your products or services, or those of NetApp so I'm a dispassionate observer of this little spat.

Really, is it the behaviour of a VP Global Marketing to engage in the sort of school yard name-calling that's been going on here? All I'm hearing is 'my stuff's better than yours and my dad could whip your dad in a fight anyday.'

You compete with NetApp and you do things differently. Leave it at that and let your customers decide what they want to buy. This sort of childish behaviour doesn't do you any favours.

[Chuck Hollis]

Hi Neil

Are you referring to my writings, or the folks who are leaving comments? I can only be responsible for what I write, and not others.

You're right -- the two companies do things differently, and customers make their choices.

One of the things that I do in this blog is offer up commentary on other vendors in this industry.

We have found that when I (or anyone else) happens to do this in regards to NetApp in particular, their employees tend to make it very personal and unprofessional in response.

Generally speaking, I hold my tongue in response, but occasionally I offer a swat or two back at them to discourage them from continuing to bully anyone who happens to criticize them or their products. By comparison, most other vendors' employees are far more professional in the banter back and forth.

You and I are different, however.

When I find objectionable material on a blog, I usually don't feel called upon to offer up scathing personal comments to the author.

-- Chuck

[Neil Fletcher]

Chuck

Fair point. Criticism accepted. I'll keep my opinions to myself in future.

Regards, Neil.

[Chuck Hollis]

Neil

Please don't misunderstand -- we all would like you to share your opinions. Just remember that there are real people here.

Thanks

-- Chuck