CX4 Announcement: Reactions and Perspectives

I was travelling yesterday, so I didn't get a chance to offer my thoughts on the CX4 announcement.

However, I do have the luxury of stepping back a bit and offering some personal perspective on what it all might mean.

And I think there's more to the story here than meets the eye.

The Basics

On the surface, it looks like yet another product announcement: EMC launches new CX4 midrange arrays.  So, what could be so special?

Well, for starters, there's the stuff you usually expect with a storage announcement: more performance (twice), more scale (twice again) and much bigger stats: drives, cache, ports, etc.

But if you're close to this stuff, you'll recognize that this isn't your ordinary let's-bump-things-up-a-bit type of refresh -- it's an entirely new architectural base that's compatible with CLARiiON that are already out there.

OK, so far so good.

And then there was the extended feature list that reads like a compendium of advanced storage features.  Things like support for enterprise flash drives.  Virtual provisioning.  Energy-saving features like drive spin-down and new low-power 1TB drives.  Modular front-end I/O that supports incorporating the next wave of storage connectivity: 8Gb FC, 10Gb ethernet, FCoE, and so on. 

That's on top of all the stuff a CLARiiON already does today.

There was also a decent amount of what I'd call "portfolio integration", like embedding RecoverPoint replication inside of the CX.  Integrated encryption and key management with PowerPath and RSA.  And more.

Finally, existing customers could easily take advantage of the new features.  Much of the new functionality is delivered in software, hence now available to earlier versions of the CLARiiON.  Before long, we'll announce the data-in-place migration feature that allows customers to use new controllers with their existing disk drives. 

And so on.

If this was an Olympic competition, I bet we'd get a "10" from the judges.

Reactions Are Interesting

Given EMC's position in the marketplace, the announcements were widely previewed (and at a fairly accurate level) in a number of publications.  Interest from the tech media was stronger than I'd have any right to expect.

I think some of the better coverage was StorageZilla, Virtual Geek who covered the VMware angle (naturally), and the good folks over at Blocks and Files. 

Some of the analysts had interesting perspectives, including Mark Peters from ESG, which offered up that -- essentially -- given EMC's stature, this is what the industry has come to expect:

"Let’s be very clear; if this CX4 product was from many other vendors, we’d probably hang out the bunting and heap on the praise. However, a lot is expected from EMC and, rather like a child that consistently performs well at school, the bar gets set ever higher. However, by concentrating on addressing user pain points and not just on providing technological advance, the CX4 looks to have cleared this higher bar.

EMC is either ‘blessing’ or leading (and either way, driving) a number of key trends; from flash drives and unifying storage platforms to driving ‘greener’ storage. Compared to the competition, the CX4 offers a very thorough and competent function set; although there are other vendors that might beat the CX4 in individual features, the new CLARiiON is overall a very solid and comprehensive offering . And winning in this particular storage game can be viewed as a decathlon—ultimately, what matters are general ability and the overall score, not the position in each individual event."

But it wasn't all positive ... as you'd expect.

CLARiiON and Symmetrix Confusion?

These product announcements get covered by a lot of people who aren't in the guts of the storage industry, and there were a number of questions along the lines of "gee, now that CLARiiON is so big, what happens to DMX?". 

Funny, when we announce smaller DMXes, the question gets reversed, e.g. "gee, now that DMX is so small, what happens to CLARiiON?".

I think people are focusing on easy-to-see specs, like disk drives, port counts, cache amounts and the like.  Well, there's more to a storage architecture than speeds and feeds.

Rather than get into an extended discussion about large global caches vs. smaller partitioned caches, or multi-controller architectures vs. dual-controller architectures, I'll leave it at this: the people who buy these products mostly know the salient differences, and know which ones play best in which environments. 

Sorry if some people can't see the distinctions.  And, yes, theoretically EMC and other vendors could do a better job in educating people in the finer points, but -- then again -- it's not really a big issue from where I sit. 

Most customers know the differences -- that's what matters.

Why Do You Have Multiple Products?

Answer: because we're an industry leader, and not all customer requirements are the same.

I love this one.  I think NetApp (as well as the rest of the smaller, single platform vendors) have done well in convincing the industry watchers that it's desirable to solve a broad range of customer requirements with a single, automagical platform. 

Desirable?  Of course!  Practical?  Hardly ...

Just like there's not a single "car" platform that addresses all markets, there's not a single "storage" platform that addresses all use cases well.  The car platform that makes a nice sports car isn't the same car platform that makes a great SUV.  Like with most things, there are tradeoffs and choices to be made.  To claim otherwise is disingenuous.

Even the most vocal proponent of this line of wishful thinking (NetApp) has had to face this technical reality, and now has multiple architectures, one example being their forthcoming VTL product which will most certainly *not* use OnTap.  And, yes, there's a distinct market of customers and use cases that want a unified storage product, one example being Celerra which does NAS, iSCSI and *real* FC all in the same platform. 

But that's a defined use case, and not all use cases. I think it's pretty obvious that the whole premise is unrealistic for most larger enterprises.

As an example, if you happen to own more than one car, I bet they're not both a Toyota Prius. 

So, What's Ahead?

I think things just got harder for all the vendors who compete in this category.  The server vendors have it a bit easier, as they can wrap servers and storage together and make it somewhat more appetizing for many customers.

But the standalone players now have a lot more to do in the "splaining" department: explaining why they don't support flash, explaining why they don't have a modular I/O front-end, explaining why they don't do spin down, or explaining why they aren't using a 64-bit architecture, or explaining why they don't have extended security features, and so on. 

That's a lot of 'splaining.

As a result, I can't help but guess we'll see far more "single feature marketing" in the the near future.  Pick your smaller vendor, pick your standalone feature -- I can't see any other option for this crowd.  And if you can't change the discussion, it's gonna be hard for you.

I also think there's a bit of acknowledgement that CLARiiON (which was an early EMC acquisition, by the way) has done far better than anyone had any right to expect at the outset.  From the early FC4700 to today's CX4, there's been a steady and predictable cadence of enhancement and extension that's a great story in itself.

And there's always more to come ...

Comments

[Martin G]

Looks very nice Chuck, but not wishing to sound like an EMC fanboy and it's relevant to an earlier blog entry; I'd like to see a CX4 certified against the v-series Filers from NetApp. I think that could be a very powerful combination in the NAS space.

And I know that there are some interesting developments within the Celerra space but it would make my life easier if every EMC product was not seen as yet further lock-in to EMC.

I'd like to see my current disk suppliers playing nice; in our case that's EMC, IBM and NetApp. I'd like to be able to put EMC disk behind filers and perhaps IBM disk behind Celerra or at least have the option to do so. I understand the reasons why EMC don't want to do this but it is becoming increasingly hard to explain to my senior management why EMC don't play and I know that I'm not the only customer who wants to do this.

But lets not take anything away from you, the CX4 looks like a very nice box and if it comes in at the price-point which is being whispered to me, it's an interesting proposition.

[Chuck Hollis]

I can't argue with your basic premise ... it'd be great if customers could mix and match freely in the IT stack, and have it all work, and all be supportable.

But we all know that's not the case: certainly for storage, and also true for other parts of the IT industry.

Right now, the storage model is that any layer has to provide support for the layers below it. Put a filer (or virtualizer) in front of someone else's storage (anyone else's) and that vendor has to be able to support the underlying pieces.

As an example, EMC sells products like RainFinity, Invista and ControlCenter that work with NetApp, HDS, IBM, etc. Before we could make that offer, we had to spend big-o bucks acquiring these competitive platforms, qualifying them, and keeping them around when a customer calls with a problem.

You don't call NetApp when RainFinity or ControlCenter or DiskXtender isn't working, you call EMC. Same for Invista with HDS and IBM. Or PowerPath. Or ... well, you get the idea.

Same logic applies in reverse to the makers of filers and virtualizers -- someone's got to own the support relationship end-to-end.

Can't argue with your logic of desirability. But I also can't see anything like this happening broadly across the storage industry anytime soon.

Thanks for writing!

[Sjon]

Chuck,

Frankly I think you'd have a better chance of getting "10s" from these judges: http://en.wikipedia.org/wiki/European_Soundmix_Show

[mgbrit]

Funny, when we announce smaller DMXes, the question gets reversed, e.g. "gee, now that DMX is so small, what happens to CLARiiON?".

Chuck, this made me laugh. I've heard the opposite, but not this. I guess I must have missed that launch announcement of the $25,000 DMX!

Joking aside, the CX4 looks as excellent as the CX3 and then some. Looking forward to getting my hands on one!

[Chuck Hollis]

Trust me on this, you get some strange questions out there.

My favorite example was a reporter who wanted to know how much a DMX weighed. I told him.

Then he asked "how much does it weigh when it's full of data?".

Well, that depends on the data type, you see, compressed data ... :-)

[Barry Whyte]

^ how much it weight with data... lol

Does a 1 weight more than a 0 ... lol

There are some weird ones out there!

[itguy]

Chuck - what's your take on the solid state drive option in Clariion? Do you think that the Tier 0 level will make sense for most users of the box? To me it the value proposition seems more obvious in the DMX... Thx for sharing your thoughts!

[Chuck Hollis]

People use CXs in all sorts of interesting situations, so I tend not to bias one way or another.

Simply put, if you find yourself striping multiple volumes, or short stroking, or both -- to obtain the IOPs you need, the enterprise flash option should be considered.

And, yes, the controller can more than hold its own with EFD performance levels in most cases.

[James3678]

How are they going to manage cooling in the CX4-960? With CX3, it's a typical front to back arrangement. With the double density in the 960, they'll be mounting the drive trays back to back, creating a heat chimney on the back side of the drive shelves. Front to back no longer works, now it's bottom to top. I guess you need to have a pretty big hole in the floor tile to push air up through the cabinet.

I also am wondering about the weight. A CX3 cabinet runs cluse to 1500 pounds. If you double the number of drives, you're going to be pushing 3000.

The flexibility in connectivity is going to be great, but I expect we would prefer a pair of 480s versus a single 960.

Also had a question about the Flare OS. Since the new platform is 64 bit, is Flare 27 going to be compatible with the CX3, or are you going to need to replace the storage processors to get the new functionaility.

[Chuck Hollis]

All great questions. And I think there are good answers. I just don't know what they are ;-)

If it's OK with you, I'll get someone online here who can answer specifically ... thanks!

[Rodan Zadeh]

Good questions.

Let me point out two key points on your remark about cooling. At the outset, on CX4 we introduced several industry’s firsts as well as many improvements over the traditional systems on environmentals. These include: low power SATA drives, adaptive cooling, virtual provisioning, and flash drive technology - to name a few. So, right off the bat employing these technologies would get you lower power usage and less heat generated as a result. Secondly, and perhaps more importantly, the ability to support more drives in the CX4 doesn’t come at price of increased footprint (i.e., bigger racks). CLARiiON remains as modular as ever. The customers don’t have to buy the all the racks and drive trays upfront to scale CX4s up to 960 drives. The arrangement of drive trays and the overall density per rack has not changed drastically on the new CLARiiON. Depending on the application and the drives deployed, the system’s weight and cooling per rack are the same or better than CX3, especially with all the innovations we did around energy efficiency.

On to your question about the new FLARE; the new 64 bit is for running on the CX4 hardware at this point.