There is always a cost/benefit analysis exercise. How much benefits do I get out from my willingness of being lock-in ? For new startups who want to tap into the Saleforce channel, it may be justifiable to use APEX. But for large enterprise who already have existing application and established customers. The lock-in concern may just be too big.
However, IaaS has an edge over the PaaS vendor. Amazon will say: "If you concern about lock-in, then don't use S3, SimpleDB, SQS API, we still have EC2 for you". I guess Google and Salesforce have nothing to say.
I haven't looked deep into Azure yet. But I have a hard time figuring out how does the conflict of interests work out here. Cloud computing is fundamentally destroying the enterprise software sales model which is the main revenue stream for Microsoft. It will be interesting to see how Microsoft is going to convince enterprises that they have better idea than Amazon, Google in running large scale web sites.
Rgds,
Ricky
From: Krishna Sankar (ksankar) <ksankar@cisco.com>
To: cloud-computing@googlegroups.com
Sent: Wednesday, December 17, 2008 11:16:48 PM
Subject: [ Cloud Computing ] Re: PaaS vs. IaaS
Stuart,
Couple of points:
a) On your #2, the scaling comes from capturing the scalability patterns and architectures in the "P" of PaaS. So rewriting your app would work (Of course, as you mention, it is a non-started; no incentive to rewrite a perfectly running app)
b) Azure is a good example of PaaS while Amazon is half way between a HaaS (or IaaS as Ricky calls it) and a PaaS.
c) Yep, Azure, for now, is in an indeterminate state – a small sub set of "cloud-able" apps would work in that environment. But this is just a small step …. The giant leap comes when the implicit patterns and architectures (of the data, control and management planes) mature and become pervasive for new applications – just like any other architectural pattern.
d) Also there is the issue of self-fulfilling prophesy. Enterprise systems never had the luxury of elasticity and so they were not written that way. Web systems, otoh, are by nature elastic and so clouds came about. And now that the architecture is out there, enterprise systems will make use of it and so future enterprise systems would need and would leverage cloud infrastructures …
e) So we really cannot judge the applicability of an elastic infrastructure in an enterprise by looking at the current apps nor can we expect that current enterprise apps would migrate to a cloud infrastructure. The sweet spot there is the utilization and that is where virtualization plays …
f) BTW, I also think there is no such thing as a cloud API – there are only cloud architectures with APIs to individual capabilities. It is not the API that is the secret sauce, it is the architecture.
Cheers
<k/>
From: cloud-computing@googlegroups.com [mailto:cloud-computing@googlegroups.com] On Behalf Of Stuart Charlton
Sent: Wednesday, December 17, 2008 9:38 PM
To: cloud-computing@googlegroups.com
Subject: [ Cloud Computing ] Re: PaaS vs IaaS
I'm curious about this theory - I call it the SSS - or, "secret scalability sauce theory". I recognize this is going to be something of a generalized argument, so please undersatnd I'm not trying to to attribute this argument to you, just that your post certainly has a familiar ring to it.
The theory goes like this:
1. Existing IT applications don't scale easily (they require knowledge)
2. PaaS infrastructure can scale easily (somehow - the knowledge needed is hidden)
3. Just rewrite your applications to PaaS!
Variants of theory include "Why can't we all be like Google?", and "Shared Nothing* Uber Alles".
(*For certain values of nothing)
My ambivalent feelings towards this theory include:
If Hybrid clouds are the bridge to the Enterprise, then is PaaS going to succeed as a black box?
- I don't really think that's what Microsoft's doing. They're allowing you to take a limited set of .NET applications - as they work today - and with some relatively minor integration tweaks, have them run in the Azure cloud. Many of the traditional problems with session state , for example, still exist, unless you punt to a pure cloud service like SSDS.
- "Rewrite" is probably a non-starter, considering how many rewrites we've had (Java/Web, SOA, etc....)
- There was a clustering debate back in the 1990's - shared nothing had many problems then, it still has many problems now for a significant proportion of enterprise IT workloads. Not to say it won't become ever more popular, I have a feeling it will. Just that it's a very big hammer looking for nails of a particular shape and size.
- Is Cloud just about fast scalability? Isn't the need to deal with bursts is more of a public Web phenomenon than an Enterprise IT workload problem?
In sum, I tend to believe PaaS works well for some classes of applications, but I am skeptical about it fitting the majority of enterprise applications.
Cheers
Stu
On 17-Dec-08, at 12:36 PM, gaberger wrote:
All I have to say is "Azure". If you look at the business model you will see that PaaS (in all its incarnations ) will develop into a number of common frameworks able to work together as easy as .NET, Java or Python. This will unleash the power of scalability never seen before.
My .02
-g
On 12/17/08 2:51 PM, "Ricky Ho" <rickyphyllis@yahoo.com> wrote:
I have some doubt about how far PaaS can go. Here is why ...
The core value of PaaS is the set of "higher level" services with extremely high scalability that the cloud vendor provides. However, there is substantial amount of risks associated to its users.
Your application is locked in a proprietary API and so it depends on the vendor's support. Imagine the situation that you build your apps on Google App Engine, or Saleforce's Force platform. And one day, they decide to focus in other business (remember that cloud computing hasn't brought major revenue stream to any of these provider companies). What are you going to do with your poor application which can't even be run in somewhere else ?
How about waiting our industry to come up with an cloud API standard ? Not mentioning it will take years to get to that point. When that has happened, I bet there will be a whole bunch of open source implementation to choose from. So instead of using the PaaS plaform, why don't you just use IaaS and run whatever software stack you like on top of it. Think about the situation if you are building your app on Amazon AWS, you pick the appropriate software stack (e.g. HBase will provide you with the equivalent storage service of Google App Engine). This way you have more choices on what API you like to bind to. Now, when Amazon decide to move out from their cloud business, you can move the application back to your inhouse environment and everything will continues to run as before.
In fact, one of the reason why IaaS usage mode is gaining much wider acceptance than PaaS is because existing applications doesn't need to be rewritten before they can be run in the cloud. IaaS provides a low barrier entry for users to get some hands on cloud experience. But PaaS requires a complete application rewrite, which is much harder to commit.
However, IaaS is not without its problem. For example, in the Amazon aws case, there are some technical issues which need to be resolved. For those who are interested, I put some technical issues of Amazon AWS here at: http://horicky.blogspot.com/2008/11/amazon-web-services.html
Rgds,
Ricky
From: Shane Brauner <shane.brauner@gmail.com>
To: cloud-computing@googlegroups.com
Sent: Wednesday, December 17, 2008 10:44:29 AM
Subject: [ Cloud Computing ] Re: Will Cloud Displace Internal IT Services For Data Centers?
In current incarnations, your major cloud options are mainly IaaS - you get virtualized hardware just like a colo. It's at the next levels up where you're going to start seeing more inroads into internal IT.
I'm coming at this from the Infrastructure / Platform / Software tri-cloud perspective. At the platform layer, you no longer need to concern yourself with any of the OS/Hardware/VM management, configuration, or administration. You focus on your code, and forget about the machines. That's going to have a big impact on internal IT.
Granted - clearly not every application is suited for this. But it's a changing world. There are aspects of business for which this is a great fit and they'll be early adopters. This will spur development of more features and functionality which will in turn allow for adoption by a broader market. It's a feedback loop.
Shane
--
Shane Brauner
http://www.10gen.com/
http://twitter.com/shanebrauner
On Wed, Dec 17, 2008 at 10:45 AM, Ricky Ho <rickyphyllis@yahoo.com> wrote:
Right ! Large enterprise is unlikely to displace their internal IT with Cloud computing.
However, they will use Cloud Computing in 2 specific ways, which is described here ...
http://horicky.blogspot.com/2008/12/does-cloud-computing-make-sense-for.html
Large enterprises requires a new suite of "management / middleware" technology which enable their applications to work in a hybrid environment (a mix of public + private cloud). Also note that the cost dynamics in public and private cloud is very different. The technology should include a cost-aware scheduler that can deploy the application components in the most cost effective way.
Rgds,
Ricky
From: "Pietrasanta, Mark" <Mark.Pietrasanta@aquilent.com>
To: "cloud-computing@googlegroups.com" <cloud-computing@googlegroups.com>
Sent: Wednesday, December 17, 2008 6:32:11 AM
Subject: [ Cloud Computing ] Re: Gartner: Will Cloud Displace Internal IT Services For Data Centers?
But again, it seems like people are missing the point:
1) Cloud Computing, at least in any of its current and pending forms, does almost nothing to reduce my internal IT staffing needs. It's no better than Co-Lo in terms of IT staffing requirements;
2) CC is more expensive than *any* of the alternatives, except in fringe cases (e.g. those with extremely volatile volume changes, and academic/research/"super computing" needs)
CC can't possibly replace internal IT services until CC *offers* some form of IT services. And the costs have to come way way down before it becomes a realistic alternative to Co-Lo (or for the small business, setting up a machine under their desk and hiring a college kid to manage it).
From: cloud-computing@googlegroups.com [mailto:cloud-computing@googlegroups.com] On Behalf Of Krishna Kurapati
Sent: Wednesday, December 17, 2008 9:18 AM
To: cloud-computing@googlegroups.com
Subject: [ Cloud Computing ] Re: Gartner: Will Cloud Displace Internal IT Services For Data Centers?
My View:
It also depends on size of organization. For SMBs, Core applications as well as non-core applications (if any) will shift to Cloud/SaaS model.
And with recent financial turmoil, many large companies became medium and medium became small :)
Any organization adopting cloud would be security/compliance, availability/accessibility.
and portability/migratability.
These requirements overweigh cost advantages depending on the vertical they play into.
Krishna Kurapati
Cloud Ventures
On Wed, Dec 17, 2008 at 3:37 AM, nagarajansankar@gmail.com <nagarajansankar@gmail.com> wrote:
Here is an interesting article that appeared yesterday.
http://www.informationweek.com/blog/main/archives/2008/12/gartner_will_cl.html?catid=cloud-computing
My take is that possibly Non-mission critical or Non- revenue
generating applications (the so called departmental applications) in
enterprises that may form about 20 to 25% of the total IT
infrastructure and services may find their way to the clouds..
Do share your thoughts...
- Sankar
http://www.linkedin.com/in/nsk007
-- <image.png>
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google
Groups "Cloud Computing" group.
To post to this group, send email to cloud-computing@googlegroups.com
To unsubscribe from this group, send email to
cloud-computing-unsubscribe@googlegroups.com
To post job listing, send email to jobs@cloudjobs.net (position title, employer and location in subject, description in message body) or visit http://www.cloudjobs.net
To submit your resume for cloud computing job bank, send it to resume@cloudjobs.net.
For more options, visit this group at
http://groups.google.ca/group/cloud-computing?hl=en?hl=en
Posting guidelines:
http://groups.google.ca/group/cloud-computing/web/frequently-asked-qu...
This group posts are licensed under a Creative Commons Attribution-Share Alike 3.0 United States License http://creativecommons.org/licenses/by-sa/3.0/us/
Group Members Meet up Calendar - http://groups.google.ca/group/cloud-computing/web/meet-up-calendar
-~----------~----~----~----~------~----~------~--~---
Posts
No comments:
Post a Comment