The industry is all abuzz with conversations about things cloud-like these days.
Sure, EMC has added our views to the discussion in terms of technology, architecture, requirements, use cases, etc.
But there's a certain element that's going to be required for this stuff to really take off.
And that's trust.
So, What Happened?
I've offered up a a post or two (or three, or four!) on some of EMC's thoughts on cloud information infrastructure. They've turned out to be pretty popular, as far as my posts go.
But, there are those that believe that we're chasing the wrong dynamic here. They're thinking commodity components, open source, advertising-driven and/or low cost business models, and so on.
My thoughts around this are -- sure, there's always room for a low-cost player or two -- but, at the end of the day, if you have something very important to you (information, applications, etc.), you're not going to hand that important stuff to any cloud unless you trust it to be there when you need it.
And then, Amazon's S3 had a little outage, according to Nick Carr's blog, which is where I picked it up.
Now, Amazon are great people, and S3 is a great service, and these things happen, but -- imagine if you were running your business on S3, and it simply went away.
No applications. No email. No customer information. No nothing.
And you weren't quite sure how long it was going to be down.
In some ways, it's almost as disruptive as a power outage, and perhaps more disruptive than the phones being down.
All you can do is hang out, drink coffee, and wait for the lights to come back on.
Engineering Trust Into Your Cloud
For me, there will be three elements to engendering "trust" in any given cloud.
First, there will need to be the right technology infrastructure: availability, performance, recoverability, and so on.
Second, the operational processes (including communicating with your customers!) will have to be absolutely honed.
Finally, I think people will respond to a brand that they already trust. Certain brands out there either already engender trust and confidence, or are entirely unknown entities.
Funny, might it end be financial institutions that succeed in this marketplace, simply because they have most-trusted brands?
And Then There's The Cost Argument
There are those that believe that, because something is cost-effective at small scale, it is similarly cost-effective at large scale. I think we're finding that's definitely not the case, especially when we're talking cloud architectures.
People's minds drift to racks of commodity servers and storage, open source distros and the like. I think they're wrong.
These technologies are not engineered to be cost-effective at massive scale. And, of course, cloud is all about massive scale. The crossover point between purpose-built and commodity will vary, of course, but, since we're talking such mind-boggling scale, I feel pretty confident saying that purpose-built technologies and architectures will most likely have advantages in long run.
And, if you think about it a bit, there are plenty of examples around you.
What's This All Mean?
I'm not entirely sure yet. I have my suspicions ...
But I think the cloud discussion has clearly re-ignited with a new passion. And everyone involved -- from producer to consumer to middleman -- realizes that there's a new pecking order coming.
There will be innovation and experimentation. Companies will make some big bets, some of which will work, many of which won't. There will be those that miss the wave entirely, and become marginalized over time.
But for IT consumers to take this stuff seriously, they're going to have to trust it.
Hi Chuck - agree 100% trust is the key issue, and a good system, even a SaaS implementation, needs some sort of contingency plan - see http://blogs.sun.com/TA/entry/web_2_0_needs_backup
My bigger question is this - As a fellow storage vendor I get your posts on Web 2.0, Active Content (not a bad name btw), new user requirements, etc. But am a bit "cloudy" on your other posts ;-)
A few years back I did a lot of research on "P2P Storage" & "Grid Storage", i.e. stringing distributed nodes together under a common namespace - you can implement this type of infrastructure as a service - is this what you are calling "Cloud Storage?"
Posted by: Taylor Allis | February 15, 2008 at 04:13 PM
Hi Taylor
The answer is "not quite". I'm not at liberty to fully divulge our thinking here, but I've dropped plenty of hints on other posts here.
Cheers!
Posted by: Chuck Hollis | February 15, 2008 at 05:09 PM
The storage cloud will be 5% technology and 95% physcology. The reason is based on the statement from the GAO architect on IT's mistakes: "Believing that IT problems are primarily attributable to technical deficiencies rather than poor management, bad processes, and managerial relationship breakdowns." I posted this list of mistakes from the GAO here: http://thinkingproblemmanagement.blogspot.com/2008/02/gao-checklist-of-it-snares-and-traps.html
There will be more IT outages like those from RIM and Amazon this week, in the future, as I have the impression that the true root causes are not being investigated.
BTW: Who is the S3 storage vendor?
Posted by: Ronald | February 16, 2008 at 12:19 AM
Hi Ronald
I'd agree with the general sentiment that IT failures are more about people than technology, especially if you include in the "people" category the failure to architect the correct environment. By that measure, all technology failures are people failures!
I don't know who sold Amazon their storage for S3. I am under the impression that it might be white-box / no-label stuff. Certainly, if one of the name brands had been used, we'd probably would have heard about it.
Thanks!
Posted by: Chuck Hollis | February 16, 2008 at 08:44 AM