More "Raiding The Roadmap"

A while back, I wrote a rather snarky post on how some vendors in this business -- when faced with challenging market conditions -- "raid the roadmap" and pre-announce all sorts of features that customers won't be able to actually buy -- let alone implement -- for a considerable amount of time.

Leaving aside for the moment the rather sensitive issues associated with revenue recognition and corporate accounting standards, one has to wonder what this new style of "pre-announcement" means for the industry -- and, more importantly, what this means for customers.

A Big (Pre) Announcement Today From My Good Friends At NetApp

NetApp's (pre)announcement today included a few bits and pieces.

First, the announced that -- sometime next year -- they'd be supporting 8Gb FC.  This announcement was in addition to their previous announcement at SNW that -- sometime next year -- they'd be supporting native FCoE as well.

OK, no surprise that they'd eventually be supporting both (duh!) but why pre-announce?  Why make it look like you could buy it today, even though you can't? 

Were there doubts that their current systems couldn't handle the new protocols?  (Not that we at EMC would ever do anything to foster those concerns with our recent CX4 announcement and its modular approach to front-end I/O, ensuring easy upgrade paths to new storage interfaces).

So, maybe the preannouncement was a direct response to, well, competitive pressures?

The second major component was enterprise flash. 

No, not the "we'll use it as cheap DRAM for cache" type of thing that they were talking about earlier in the year, but in the more traditional "we'll provide drive shelfs with enterprise flash for persistent storage of data".

Huh.  So despite all the FUD from them and everyone else on EMC doing this first in the DMX, and more recently on the CX4, I guess it wasn't such a bad idea after all, was it?

Except that you can't buy it today from them -- maybe sometime next year, if I read the (pre)announcement correctly.

Now, knowing what I do about how a storage array wants to use flash, and knowing how NetApp and others have designed their arrays around spindle-randomizing, this should be Really Interesting to see how they resolve these two different agendas.

If you're interested in this topic, go see my earlier posts on data placement and why this is a very important consideration when considering enterprise flash drives.

Finally, think about this one for a moment.

If you put really fast pipes on one end of a storage controller, and really fast flash drives on the other end, you've got a good case for a major storage controller overhaul to modern x64 multicore technology (as seen in the CX4).

Hmmm, didn't see that one in the (pre)announcement either. 

[update: Alex McDonald over at NTAP appears to be implying that all NetApp products have been full 64-bit since the mid 1990s.  But, based on my interactions with Val, you've got to pin these guys down with very specific questions.  I'm not close enough to comment on this -- can someone verify that all current NetApp FAS products are true 64-bit architectures?]

It's one thing to (pre)announce a feature, it's another thing to be able to extract the required performance from the (pre)announced feature.  And I'm betting that we'll see a major controller upgrade from them (and just about anyone else that goes down this path) before too long.

We know of what we speak -- that's what the CX4 was all about.

This sort of inevitable controller upgrade to support the new (pre)announced technologies kind of gets in the way of that "your investments today are protected" message, doesn't it?

Also mentioned are a few "new" offerings around remote support -- something we at EMC understand pretty well, I'd offer.  No mention, however, about providing one-time security credentials, encrypted communications, anti-spoofing measures and tamper-proof audit logs for remote support activities, so I guess they're not recommending that customers use this stuff in security-conscious environments.

If I were over at NetApp, and (pre)announcing something, I'd play to strengths, and (pre)announce something very practical, like drive spin-down for all those big filers with infrequently accessed data. 

But no, they're chasing the performance vector.  I wish them luck with this one.

What Does This Mean For Customers?

On one hand, any "statement of direction" is a good thing for customers, up to a point. 

If there's a particular technology you're interested in, it's nice to know that a given vendor agrees with you that it's interesting, and is willing to invest in supporting it.

But when we're talking something that's not going to be delivered for 6 months, or 12 months, or maybe longer -- well, a lot can happen along the way.  Especially if it's a core capability that's essential to your planning processes.

And I hate to see customers hung out to dry waiting for something that comes along later than planned, or has a few "unexpected considerations" at final GA, or -- worst -- doesn't end up getting done.

I do take exception, however, to the practice of making future capabilities appear as though "they're here today!". 

Because, well, they're not.

And, Finally, This One Takes The Cake

I'm sorry, this last one triggered my cynicism circuits into overload:

"NetApp to Lead Industry Discussions Around the Future of Solid-State Storage and Parallel NFS in November 2008"

Of course, it's great that NetApp is participating in various industry forums and standards groups -- the more the merrier.  EMC does the same, but we don't think that sort of thing usually warrants an Official Press Release, unless it's a pretty pivotal standard for the industry (e.g. CMIS and XAM being recent examples).

Now, please appreciate -- just for a moment -- how this sort of hype plays here at EMC. 

Let's see, they're at least a year behind EMC's pace in introducing solid-state storage into shipping products, never mind all the FUD NetApp shared with the industry early on.   The best they can do is pre-announce something for perhaps next year, and maybe it will work OK.

And, interestingly enough, EMC has been shipping a pre-standard pNFS implementation (known as MPFS) for several years, and we have all sorts of interesting and practical insights into where it fits best, customer examples, implementation issues, scalability issues, etc.  Nothing like that in the NetApp portfolio today.  Want to experience what pNFS-type solutions might be able to do for you?  For that, you'll have to call EMC today.

So, will someone please tell me, just what might qualify them to "lead the discussion" on either of these topics?

Yuck.  One of the worst forms of marketing hype, in my personal book.

What do you think?  Courteous comments welcome, as always ...

Comments

[Chris M Evans]

Chuck, I have to admit I almost choked on my coffee when I read that "lead industry discussions around solid state" announcement.

I'd agree that EMC tend not to announce products until they're officially ready (or close to, in the case of TP for DMX-4, from memory) however would you not agree that EMC does use rumour to pre-announce products - Hulk/Maui for instance?

[Chuck Hollis]

Hi Chris --

The now-public discussion around Maui was hardly intentional.

I don't want to share how it all started, but let's just say that a bit of information slipped out, and then we all have to dance around it until the product is officially announced.

Nobody really planned it that way, it just happened.

[Chris M Evans]

Chuck, sorry, that wasn't a dig at you and if it came across like that I apologise. No, I was referring to Mr Tucci announcing it last November.

[Chuck Hollis]

No offense taken, Chris! Like I said, inadvertent disclosure comes from all sorts of sources :-)

[Alex McDonald]

Chuck, there's so much to refute in your latest snark, I was forced to blog in case I burst the comment box.

http://blogs.netapp.com/shadeofblue/2008/11/hairy-canary-ov.html

[Joshua Konkle]

Check out the blog on presenting pNFS at SNW post here:
http://blogs.netapp.com/engineeringapps/2008/10/snw-recap-pnfs.html

I was leading the talk (tongue-cheek) and David Black, Distinguished Engineer at EMC was supporting the effort from the audience. We're now working together to broadly communicate the value of a standardized pNFS client and server environment.

We took the lead on presenting it there and are glad to have your firms support moving forward.