Share what you know with millions of people
Focus is the best place to turn what you know into remarkable content
0
When it comes to cloud computing are you on Team Ellison (Oracle) or Team Benioff (Salesforce.com)?
Last week at Oracle OpenWorld the focus of the whole event was on cloud computing. From Larry Ellison’s opening keynote, to Salesforce.com founder Marc Benioff’s session, all the way to Ellison’s ending session, the cloud was the star. But the funny thing is the very different approaches and philosophies companies are taking with developing and utilizing cloud computing services.
But the only thing Ellison and Benioff seem to agree on when it comes to the cloud is how to spell it. Benioff described the cloud as being pay-as-u-go, elastic, real-time, 5x faster, half the cost, and multi-tenant. And while Ellison agrees with the elasticity, speed and cost elements, he definitely didn’t agree with the multi-tenancy part – of which he said was a horrible idea from a data security perspective. And Ellison also emphasizes the importance of private clouds - where companies can manage and support cloud environments for their internal customers.
So which cloud philosophy and approach do you more closely align with? Ellison and Oracle? Benioff and Salesforce.com? Or somewhere in between?
Events
- Dos and Don'ts of Small Business Marketing May 29 @ 11 am PT
- Lead Nurturing 202: The Next Generation May 31 @ 11 am PT
- The Tricks to Paid Media June 6 @ 11 am PT
- Display Advertising for Brand Awareness June 20 @ 11 am PT
Related Activity
7 Answers
There are two things about this argument that don't work in Larry's favor -- say what you may about security in multi-tenancy (which is not an issue outside of Oracle's mind) but you can not have a cloud application without multi-tenancy -- it would not be a cloud (proper definitions, I know -- nitpicking).
The other part of his argument that does not work well, he calls it private cloud (at which point the issue of multi-tenancy ceases to be an issue -- if you own the "cloud" and all other components - why bother with multi-tenancy? just distribute the load among your components, far easier and cheaper and faster). Oh, yeah -- just like there is not such a thing as a single-tenant cloud app, there is no such thing as a private cloud.
The former is called client-server app, the latter is called SOA-style IT architecture.
Now, here is where Benioff's arguments begin to fall apart -- there are three parts to a cloud implementation: IaaS, PaaS, and SaaS (infrastructure, platform, and software). I am not certain why Salesforce insists in leaving PaaS out of the discussions when they have, arguably, the best PaaS and more complete layer of the two. Alas, be it -- Salesforce offers only SaaS (and PaaS in disguise), but no IaaS. Guess who has all three layers?
Yup, Mr Ellison's cloud-in-a-box (well, grossly implemented -- but they are there...and I made a promise I would not make fun of a cloud-in-a-box).
So, to assign points -- Marc gets points for closer-to-reality marketing, Larry gets geek points for closer-to-technology.
Winner?
Amazon and their Grid, IaaS/PaaS rental model (not a cloud) that got more mentions as the one to emulate from both of them than any other product.
Hi Brent,
Good question on a topic that is hot and will be with us for a long time. I think it's important to listen your customer/constituency and deliver what they need/want. At this point the interest by Nimble END Users/VARS has been minimal for a local data store/private cloud version. We are prepared technically to flip a switch to enable private/local data stores if we get sufficient demand for it.
I wonder if Mr. Ellison's (Oracle's) position is based on the need to sell/justify all the local Oracle database servers installed at his customers sites?
I believe both positions are valid and that vendors will have to serve their customers needs in the end. I wonder what their position on the cloud will be after a large hosted data center burps, gets hacked or is otherwise unavailable? It's interesting how people opinions change after a disaster. I do think we will all wake up one day and be shocked how little privacy we all have left. This may result in a temporary backlash against the cloud. In the end I do believe that expediency and cost will prevail and the majority of us will live in the clouds,
Thanks for the great question.
Best,
Jon Ferrara
Brent,
Since the core of the differences stated have to do with tenancy, I will focus on that for the moment. There are good arguments to be made for Multi-Tenant and for Multi-Instant (or copies of single tenant). The key to the question is who is the primary beneficiary of the technology or infrastructure chosen. Salesforce.com chose Multi-Tenant because it was the best fit, at the time (over 10 years ago now). It may still be the best fit (I do not want to start a long debate here). If your provider can guarantee security, performance, etc.,... the question is, does it really matter?
Tenancy is a matter of perspective. All SaaS applications are multi-tenant depending upon the layer you are considering. For example, putting two Virtual Machines on a single physical machine is really Multi-Tenant at the level of the operating system, but not at the application layer, where it would be multi-instant. As long as the application provider can maintain code and distribute changes in a way that upgrades are not an issue, it really does not matter to the user of the application - as long as it just works.
Mitch
Hi Brent,
I think the answer is: it depends. No big surprise, I know. The question really comes down to what is implemented on the cloud. Most implementations of cloud computing will involve processing not data storage. I need to scale quickly to handle some increased demand for transactions. This is much leas expensive then building a server farm for the hugest possible demand or turnning away customers with: Please try again later. This type of computing is mostly stateless and what ever data is used will stay mostly in memory and not on disk. It should be easy for developers (and cloud computing providers) to secure that sort of environment on a multi-tenant cloud.
There is a data implementation on a cloud which replaces databases completely. This is called a distributed hash table. You can understand why there would be a bias against this type of cloud computing at Oracle. There are benefits to a DHT inside a firewall but the real benefits come from the implementation's non reliance on data location. It allows you to share data with anyone anywhere and optimize for actual needs. For something on that scale you can do either behind the firewall of multi-tenant but the difference is that you still have multi-location; either multi-company or multi-datacenter. You end up with the same security concerns.
In short (I know, too late). If you are storing data for one company in one location then just use Oracle. If you are storing a lot of data in multiple locations you can implement a DHT using multiple companies hardware or use a multi-tenant cloud. If you are load balancing computing power then use a multi-tenant cloud implemented securely by your developers.
While I really like what Oracle is up to, I think the use of 'cloud' in this context is a little bandwagon-ish. The term is pretty ambiguous in general, but I think Benioff's use is closer to what technologists have come to accept.
In Oracle's case, as it relates to the 'private cloud', he's talking about scalable, highly-available network infrastructure. As soon as you put it behind the firewall, I'm more inclined to call it cool network tools than being anything cloud related.
Larry also knows that multi-tenant, especially if tackled by a company like Oracle, is not something that should be feared. He's making a case for traditional IT big-box infrastructure, while trying to appeal to buyers how are turned on by the cloud.
Ellison has furthered the argument that 'cloud' is a marketing term, not a technology term. Local servers by any other name...are still local servers.
Brent,
I agree with @Jon that the primary reason behind Larry's position is that now he's got a boatload of software and hardware to sell so it's not likely he'll support concepts that favor a public Cloud with a tightly integrated PaaS + IaaS platform as SFDC has.
I think that @ekolsky has hit most of the main points.
I will disagree with @Mitch however. "As long as it works" is not the measuring stick that one should use. You can get things to "work" to a point spinning up multiple single-instances, but if you want to scale really big and maintain profitablity (and even see it increase as you bring on more and more customers), multi-tenancy is a pretty important concept. I actually just had part 1 of 2 published on the subject on a travel industry site called Tnooz. Please check it out. Tomorrow is the "Top 10" list of benefits of a properly architected product. http://www.tnooz.com/2010/09/28/how-to/part-one-of-two-understanding-saas-and...
Brent, Jon's making THE point here: Cloud will be used to please your customers, present ones first of course LOL
I've come to the conclusion that we shouldn't be so strict about what I call Pure Cloud - we all want a piece of Cloud. My http://www.martijnlinssen.com/2010/06/cloud-field-social-way-to-future.html explains that a bit less concise
Let's be honest: 1) traditional vendors and SI will croak if the only Cloud they're allowed to do is Pure Cloud, 2) traditional enterprises aren't ready for Pure Cloud by far
Answer This Question