"Vertically integrating the stack always fails," asserts Dave Stevens, Brocade's chief technology officer. He disagrees with the bosses of Cisco, EMC, HP, Oracle and VMware, though NetApp will love him. The Brocade pitch is that you need best of breed products at each stack layer: system software; hypervisor, server, storage and …
Proprietary vertically integrated stacks
In a conversation, Dave Stephens said he meant proprietary vertically integrated IT stacks, meaning the HP and Oracle ones but not the VCE's Vblock. He said that the IT industry was unlike other industries where proprietary vertical integration can prevail, such as , I think, automobiles and airplane manufacture. He did say that Cisco was trying to own a proprietary stack insofar as its CNAs. UCS servers and Nexus switches were concerned. There, originally, Brocade, Emulex and QLogic CNAs were supported as well as Cisco's CNA but now Cisco is mostly shipping its own CNAs based on its Palo chipset.
A company which can't fix broken capacitors ? Enterprise ? Absolutely "Best Of Breed". No more questions.
Can I have a full-page FAIL with that ?
Dell is very much a part of HPC
Dell is pretty much identical to HP in the x86 market, don't judge them only by their junk desktops and desktop printers. Remember that it wasn't too many years ago that both HP and Dell were known for superior quality and service, now they're both known for a race to the bottom in price and quality in consumer & small biz, but their server lines & support are almost indistinguishable, and both are gaining traction in the mainframe and HPC biz as well (obviously HP is much longer in mainframes; Dell is throwing an enormous amount of money and freshly-plucked engineering talent at breaking in because the margins are so huge).
Not a Dell fanboi, but know your enemies at least as well as you know your friends.
Not entirely disinterested viewpoint...
As Mandy Rice Davies (almost) said, "he would say that, wouldn't he". Hardly a disinterested party.
In some cases vertically integrated stacks work very well. Just look at what Apple have based their entire business model on. IBM did much the same, with great success, until the competition authorities forced competition and interworking into the mainframe market. We saw something similar with Microsoft who (games consoles apart) did, at least, stay clear of the hardware side. We now have Oracle building a very strong position which now covers more layers than any IT company before.
I know it all makes sense to some people (maybe many people), but it was so much gibberish to me. What the hell does "vertically integrated IT stacks are dying in the compute space" mean? Is it important? If you strip away the jargon, is there anything meaningful there?
Maybe I'm just too old. I think I'll ignore any article that uses 'stack' as a buzz-word, rather than referring to a 'first in, last out' data structure. That might do it.
Driven to drink
I have great time for Brocade, they release brilliant products (I recently saw a Silkworm 12000 that had an uptime of 8 and a half years), and it is always good to see an industry giant like Cisco fail to dominate an area which they think they should.
But I'd have to agree with JustaKOS. All this cloud stuff is doing my nut in, it's just another way of selling storage, servers, networking, OS and middleware. Same as it has ever been. EMC's "journey to the private cloud"? I've seen one, it's called a data centre.
a "vertical integrated stack" is something like this: On the bottom you have hardware designed, made and maintained by vendor X. Based on that, there is an operating system developed, integrated and maintained into said hardware by vendor X. On top of that, there is a database, developed, integrated, maintained by vendor X. On top of that the same story with the application server. On top of that... the ERP system from vendor X.
These "stacks" have the advantage that system integration/problem rectificiation costs are typically much lower than a "best of breed" approach were you take the hw, os, database, app server and ERP from a different vendor each.
On the other hand, vertical integration breeds tons of inefficiency, bureaucracy, huge headcount and excessive cost. Look at IBM in the early 1990s and you will understand.
Oracle was highly successful with a different approach: Selling the Oracle RDBMS for all conceivable hardware/operating system platforms. From VMS to NCR via Unix, so to speak. They ate lots of business from IBM, who were vertically integrated then (they even made their own RAM and harddisks).
Now Oracle acquired SUN and they need to spin this. So they praise the "vertical blabla". We will see how this "pans out". I am highly sceptic. Intel already killed SUN and the Oracle hw biz ist next on the list.
Integration is in the eye of the beholder
Integration will win because it drives out cost in most businesses.
Unless you are at a scale where only "rolling your own" makes sense, you will simply choose how much "integration" you want to buy, and how much choice you are willing to give up in exchange for simplicity. Oracle contends that "the flower grown in the walled garden will always be more beautiful than the flowers in the meadow". As long as it is an Oracle Rose.
EMC, NetApp, Cisco, Brocade and others want you to be able to pick the flowers you want to grow in their gardens.
Let the games begin.