Do you see Applications breaking further into the commercial space similar to salesforce.com's business model? How does Cloud Computing differ from an Application Service Provider business model? Do we see other application such as ERP and other such applications migrating to the Cloud? How important will value added professional services and outsourcing services play in the maturation of Cloud Computing for businesses? What about the need for business to have customization and data integration to other systems?
I really like what you are saying here, specially the part where you tell people how it is about optimizing software design and to take into consideration that hardware will eventually fail. It is true that running on reliable servers is expensive.
Great post, the only thing I did not like is that at the last part you make the post sound so much like a big Ad for App Engine, and Gmail.
Rajen, the issue for a number of us enterprise customers interested in private clouds is neither specifically the technology nor the cost/unit. Rather, for organizations like Universities (and health care, some government), there is both a regulatory and a privacy requirement. Private clouds appear to be a reasonable approach (with less risk) until such time as you and others in the fully flat cloud architecture allow us to layer light-weight and virtualized requirements around our business needs (in our case student and patient privacy)
There exist major obstacles including service level and security/privacy concerns.
I have yet to find any vendors providing satisfactory answers to these common questions asked by many enterprises: - Can you guarantee 99.XXX availablity (using MY definition of downtime) for the price of Y? - Can I have a portion of the infrastructure located in my building, so that in the worst case situation I don't lose everything and still have a down-scaled system under my control? - My apps make use of some sensitive data. How can you guarantee only me can access those data? Do you bear the risk and cost of security breach caused by your fault?
I read the McKinsey report with great interest but was thought that the their conclusion was lacking in a complete analysis, namely the vacancy of application outsourcing and innovation. Thanks for jumping into the fray!
Transferring the world's computing resources into the cloud is going to drastically increase the need for high speed, reliable and distributed bandwidth. Currently, much of the world including many parts of the US are lacking in broadband infrastructure. Is this a big concern for Google and are there things that Google is doing to solve the issue?
If I were starting a business from scratch, the cloud would be an appealing alternative to internally hosted infrastructure.
But as an enterprise IT manager, there are significant barriers to embracing the cloud. Most corporate IT departments manage large ERP and CRM applications, as well as home-grown legacy applications specific to their businesses. The value being created by IT projects today is not in these core functions, but in applications which integrate data and functions between them. For example, customer self-serve apps which leverage inventory, order management and CRM data. We are embracing SOA, MDM and ESB technologies to deliver these capabilities.
To leverage the cloud in a meaningful way, corporate IT managers will have to migrate applications from their internal infrastructures - it won't be possible to move everything at once, and software providers won't all support cloud-based implementations or SaaS implementations in the immediate future.
Thus, integration becomes the real challenge for most corporate IT managers today, and the usage patterns blending cloud-hosted and internally-hosted applications and data are not mature. How will I manage latency and reliability of transactions in which some of the services are hosted internally and some in the cloud? Will my ESB strategy support it? What about security and privacy? In this age of concern over customer data privacy, how can I assure my customers, shareholders and Board that I have control over customer data that is hosted in the cloud?
Until this class of challenges is addressed, corporate IT managers will not embrace the cloud for large, important parts of their portfolios. It will remain a sandbox for innovation only.
What about data security? I'd love to move to cloud computing in the future, but I run an online banking site. It is unlikely our regulators and auditors will be comfortable with sensitive financial data in the cloud.
Great article. Though I do want to point out that the viewpoint is one from only a Software as a Service and Platform as a Service perspective. I think that an organization who wishes to have a private cloud for the purpose of Infrastructure as a Cloud may be justified in doing so if most of the applications they are running are small in nature and only require a slice of a server.
Relitivly new to cloud computing but its a technology we are looking to bring into our high school. one laptop per kid all run from the cloud computing model.
Rajen, I concur. By focusing solely on the cloud's infrastructure layer, the McKinsey report perhaps missed out on the opportunity to shine the light on other aspects like the platform (infrastructure software) and applications (SaaS). Your comments around corporations being able to innovate better by relying on the cloud are right on.
Joseph Lam, you're right, but there are counter-questions:
- Can your own data centers guarantee 99.XXX availablity (using YOUR definition of downtime) for the price of Y?
- Can you have a portion of the infrastructure located outside your own building, so that in the worst case situation you don't lose everything and still have a down-scaled system able to function?
- Your apps make use of some sensitive data. Can your data centers guarantee that only you can access those data? Do they bear the risk and cost of security breaches caused by their fault?
Brandon, the notion that you control people (or computers) just because you pay their salaries (or their costs) is naive. Some of your employees don't care what you want, and some are probably working against you, and all of them are up against determined attackers with more resources (in the form of pwned Windows boxes) than you can possibly muster.
In short, the point is not to compare actual clouds with an idealized version of self-managed data centers, but to compare actual clouds with actual data centers. Neither is going to be perfect at any price, and there will always be disasters that nothing can make up for.
(Disclaimer: I work for, but don't speak for, Google.)
Great post. Cloud computing is great conceptually, but there is still a lot of FUD out there about what it is and how it works exactly. Posts like these are a great way to help influence the IT decision makers. Keep 'em coming!
Also, just a quick "Ditto" on Lev and Joseph's remarks on security and regulatory compliance. I'm currently working on a healthcare app, and we have some regulatory checkbox items that we must check off before our customers will buy. We will likely go with a rack space provider (at higher cost and less reliability) simply because that will allow us to say we are compliant. BUT we would really like to go with a cloud solution, because we can see the huge benefits.
Saying you don't need private data centers when you have scalable clouds is like saying you don't need your laptop when you have Google Apps. Purists have been preaching that model for years but you know we haven't reached that point yet.
For another 10 years, it is safe to assume that both models will be around. And IT guys will be arguing over candidates that make sense to outsource to (public) cloud and those that must be retained on private data centers.
I'm really interested in the notion of cloud computing, particularly as a way of distributing and maintaining user end software. But as a business I'd be very concerned about control of my data. I don't need to be a bank to have information I don't want shared. Outsourcing your data centre is one thing, dispersing my data into the cloud is quite another.
What guarantees do I have that my data is secure and what do you do to help me maintain regulatory compliance?
But how can you explain Amazon using XEN underneath to offer applications as a service? Do you believe that virtualization is needed in this case or not ?
The use of the word scale and scaling borders on illiteracy , it is meaningleas to those without your insider knowledge , complete technobabble, and that coming from someoneone who holds and MCSE and a PHD in Japanese.. how do you "leverage a cloud"
Do you see Applications breaking further into the commercial space similar to salesforce.com's business model? How does Cloud Computing differ from an Application Service Provider business model? Do we see other application such as ERP and other such applications migrating to the Cloud? How important will value added professional services and outsourcing services play in the maturation of Cloud Computing for businesses? What about the need for business to have customization and data integration to other systems?
ReplyDeleteI really like what you are saying here, specially the part where you tell people how it is about optimizing software design and to take into consideration that hardware will eventually fail. It is true that running on reliable servers is expensive.
ReplyDeleteGreat post, the only thing I did not like is that at the last part you make the post sound so much like a big Ad for App Engine, and Gmail.
Rajen, the issue for a number of us enterprise customers interested in private clouds is neither specifically the technology nor the cost/unit. Rather, for organizations like Universities (and health care, some government), there is both a regulatory and a privacy requirement. Private clouds appear to be a reasonable approach (with less risk) until such time as you and others in the fully flat cloud architecture allow us to layer light-weight and virtualized requirements around our business needs (in our case student and patient privacy)
ReplyDeleteThere exist major obstacles including service level and security/privacy concerns.
ReplyDeleteI have yet to find any vendors providing satisfactory answers to these common questions asked by many enterprises:
- Can you guarantee 99.XXX availablity (using MY definition of downtime) for the price of Y?
- Can I have a portion of the infrastructure located in my building, so that in the worst case situation I don't lose everything and still have a down-scaled system under my control?
- My apps make use of some sensitive data. How can you guarantee only me can access those data? Do you bear the risk and cost of security breach caused by your fault?
Thanks for a great read!
ReplyDeleteI read the McKinsey report with great interest but was thought that the their conclusion was lacking in a complete analysis, namely the vacancy of application outsourcing and innovation. Thanks for jumping into the fray!
ReplyDeleteThat was really informative and interesting. Thanks.
ReplyDeleteTransferring the world's computing resources into the cloud is going to drastically increase the need for high speed, reliable and distributed bandwidth. Currently, much of the world including many parts of the US are lacking in broadband infrastructure. Is this a big concern for Google and are there things that Google is doing to solve the issue?
ReplyDeleteCloud computing definitely the future computing model. It is really a very cost effective way & convenient & powerful.
ReplyDeletegood post
ReplyDeleteIf I were starting a business from scratch, the cloud would be an appealing alternative to internally hosted infrastructure.
ReplyDeleteBut as an enterprise IT manager, there are significant barriers to embracing the cloud. Most corporate IT departments manage large ERP and CRM applications, as well as home-grown legacy applications specific to their businesses. The value being created by IT projects today is not in these core functions, but in applications which integrate data and functions between them. For example, customer self-serve apps which leverage inventory, order management and CRM data. We are embracing SOA, MDM and ESB technologies to deliver these capabilities.
To leverage the cloud in a meaningful way, corporate IT managers will have to migrate applications from their internal infrastructures - it won't be possible to move everything at once, and software providers won't all support cloud-based implementations or SaaS implementations in the immediate future.
Thus, integration becomes the real challenge for most corporate IT managers today, and the usage patterns blending cloud-hosted and internally-hosted applications and data are not mature. How will I manage latency and reliability of transactions in which some of the services are hosted internally and some in the cloud? Will my ESB strategy support it? What about security and privacy? In this age of concern over customer data privacy, how can I assure my customers, shareholders and Board that I have control over customer data that is hosted in the cloud?
Until this class of challenges is addressed, corporate IT managers will not embrace the cloud for large, important parts of their portfolios. It will remain a sandbox for innovation only.
thanks.... its hard to explain cloud computing to muggles
ReplyDeleteWhat about data security? I'd love to move to cloud computing in the future, but I run an online banking site. It is unlikely our regulators and auditors will be comfortable with sensitive financial data in the cloud.
ReplyDeleteGreat article. Though I do want to point out that the viewpoint is one from only a Software as a Service and Platform as a Service perspective. I think that an organization who wishes to have a private cloud for the purpose of Infrastructure as a Cloud may be justified in doing so if most of the applications they are running are small in nature and only require a slice of a server.
ReplyDeleteRelitivly new to cloud computing but its a technology we are looking to bring into our high school. one laptop per kid all run from the cloud computing model.
ReplyDeleteExcellent points, but the one benefit of private virtualization is control. You control your own data and processing, not an external 3rd party.
ReplyDeleteRajen, I concur. By focusing solely on the cloud's infrastructure layer, the McKinsey report perhaps missed out on the opportunity to shine the light on other aspects like the platform (infrastructure software) and applications (SaaS). Your comments around corporations being able to innovate better by relying on the cloud are right on.
ReplyDeleteI think the best solution is to combine cloud and control .
ReplyDeleteJoseph Lam, you're right, but there are counter-questions:
ReplyDelete- Can your own data centers guarantee 99.XXX availablity (using YOUR definition of downtime) for the price of Y?
- Can you have a portion of the infrastructure located outside your own building, so that in the worst case situation you don't lose everything and still have a down-scaled system able to function?
- Your apps make use of some sensitive data. Can your data centers guarantee that only you can access those data? Do they bear the risk and cost of security breaches caused by their fault?
Brandon, the notion that you control people (or computers) just because you pay their salaries (or their costs) is naive. Some of your employees don't care what you want, and some are probably working against you, and all of them are up against determined attackers with more resources (in the form of pwned Windows boxes) than you can possibly muster.
In short, the point is not to compare actual clouds with an idealized version of self-managed data centers, but to compare actual clouds with actual data centers. Neither is going to be perfect at any price, and there will always be disasters that nothing can make up for.
(Disclaimer: I work for, but don't speak for, Google.)
Great post. Cloud computing is great conceptually, but there is still a lot of FUD out there about what it is and how it works exactly. Posts like these are a great way to help influence the IT decision makers. Keep 'em coming!
ReplyDeleteAlso, just a quick "Ditto" on Lev and Joseph's remarks on security and regulatory compliance. I'm currently working on a healthcare app, and we have some regulatory checkbox items that we must check off before our customers will buy. We will likely go with a rack space provider (at higher cost and less reliability) simply because that will allow us to say we are compliant. BUT we would really like to go with a cloud solution, because we can see the huge benefits.
Saying you don't need private data centers when you have scalable clouds is like saying you don't need your laptop when you have Google Apps. Purists have been preaching that model for years but you know we haven't reached that point yet.
ReplyDeleteFor another 10 years, it is safe to assume that both models will be around. And IT guys will be arguing over candidates that make sense to outsource to (public) cloud and those that must be retained on private data centers.
I'm really interested in the notion of cloud computing, particularly as a way of distributing and maintaining user end software. But as a business I'd be very concerned about control of my data. I don't need to be a bank to have information I don't want shared. Outsourcing your data centre is one thing, dispersing my data into the cloud is quite another.
ReplyDeleteWhat guarantees do I have that my data is secure and what do you do to help me maintain regulatory compliance?
Nice Post...
ReplyDeleteBut how can you explain Amazon using XEN underneath to offer applications as a service? Do you believe that virtualization is needed in this case or not ?
ReplyDeleteThe use of the word scale and scaling borders on illiteracy , it is meaningleas to those without your insider knowledge , complete technobabble, and that coming from someoneone who holds and MCSE and a PHD in Japanese.. how do you "leverage a cloud"
ReplyDeletecomplete non English pseudospeak .. I despair ...
regards
Paul Burns