Cisco Blogs

Cisco Point of View on Cloud Computing

- July 8, 2009 - 31 Comments

I had the opportunity last week to deliver a keynote at Cisco Live in front of ten thousand people at the Moscone Center in San Francisco and another three thousand who joined virtually. I extend my sincere thanks to all those who participated at the event. If you missed the keynote, you can watch it here.As part of my keynote, we presented a demo of private cloud computing. We showed how a group of state-government agencies and state universities could use the private cloud model to pool their IT resources into a shared compute platform, thus making their IT environment dramatically more flexible, dynamic and efficient. To go straight to a replay of the private cloud demo we showed at Cisco Live on July 1, please click here.The specific “use case” we presented demonstrated several capabilities that are already available from Cisco today. For example, the ability to add extra compute capacity in minutes using a Unified Computing System ― versus the weeks or months it takes to add physical servers manually. We also demonstrated the ability to move IT resources from one data center location to another rapidly, in order to avoid a natural disaster (say, a hurricane or tornado) or to take advantage of cleaner, lower-cost power. The first step in building a private cloud is to virtualize your IT infrastructure, because virtualization provides the foundation for private cloud computing. In essence, private clouds combine a cloud operating system with Cisco’s cloud internetworking technology portfolio to link enterprise and service provider resources into a single enterprise-managed cloud environment. This cloud is then available to any device, any where via standard TCP/IP networking technologies. Importantly, the cloud gives IT the ability to reach out and leverage the resources of cloud service providers. Private Clouds fundamentally change the dynamic between IT and the business, in the following ways:- They change the IT consumption model- Enable IT to deliver business value more quickly- Reduce inefficiencies between the business and IT- Increase the rate of business innovationOur long-term vision at Cisco is to help build what we call the “Inter-Cloud,” a future state in which federation will occur and CIOs will be able to dynamically and securely move large workloads from one service provider’s cloud to another. This capability will give CIOs choice and flexibility on a scale that’s never been possible before. We believe the Inter-Cloud will change the economics of IT and will drive a new wave of innovation and investment similar to what we saw with the Internet in the mid-1990s. I am very interested to get your feedback and thoughts on Cisco’s approach to private cloud computing. Please submit your questions to me, either here on the platform blog or via Twitter. At the end of this week I will post answers here to your top

In an effort to keep conversations fresh, Cisco Blogs closes comments after 60 days. Please visit the Cisco Blogs hub page for the latest content.


  1. When you are in uncomfortable position and have got no cash to go out from that point, you will have to take the loans. Just because it would aid you emphatically. I take sba loan every time I need and feel myself fine just because of that."

  2. Clouds seem to be way more sophisticated than those lack boxes"", remember those? Clouds will usher in an efficient pay-as-you-go service but will end up having to store/archive all kinds of work histories! And maybe redundant copies of everything to cover assurances of access. But what about liabilities? Just what kind of support can a 'customer' expect if proprietory info is tapped for outside uses? - hijacked!"

  3. Padmasree, your follow-up comments regarding User Experience"" and ""Trust"" are spot on."

  4. Link to Cloud Papers referenced above:

  5. Major EA vendors will go kicking and screaming until they figure out the business model shift.

  6. Its about explaining the process and the ROI to the CIO/Manager level IT professionals in the companies.Also, its the fear for going something new during the low economy that is keeping the organisations away for moving into cloud computing.

  7. Though issues such as security and reliability are valid, the real barriers I see --1/ The inability for IT folk to give up control and use vendor managed services.2/ Most enterprise-wide applications (like ERPs) are not able to leverage the internet nor the cloud.

  8. I think that the speed and effectiveness of Cloud Computing is definitely one of the key factors that make it so valuable to the enterprise especially. Hours can add up from administrative time, so the cloud eases the burden."

  9. Once again, great dialog here! Some very insightful comments on the barriers to enterprise adoption of the cloud model. In many ways, the central issue we are all pointing to is TRUST in the cloud. My colleague James Urquhart has written extensively on this. In order for enterprises to gain the advantages of cloud computing, we need to establish a trusted approach to the cloud. Because without trust, all the economics of cloud computing become irrelevant and beside the point. Trust in the cloud really centers on four core concepts:1. Security: Traditional issues around data and resource access control, encryption and incident detection2. Control: The ability of the enterprise to directly manage how and where data and software is deployed, used and destroyed3. Service-Level Management: The definition, contracting and enforcement of service level agreements between a variety of parties4. Compliance: Conformance with required regulatory, legal and general industry requirements (such as PCI, HIPAA and Sarbanes-Oxley) We believe the network can uniquely address these issues. Establishing a trusted approach to private clouds represents a significant area of focus for us and our partner community."

  10. In response to the question on twitter regarding the barriers for large enterprises moving to cloud computing, these are nothing like the barriers perceived by small companies - especially here in the UK. I spend most of my time trying to convince, for business, not technical reasons, CEOs and CIOs to at least look at the potential opportunities. Most perceive the main barrier to be the lack of control (really they are saying the lack of ownership when you mention rental or lease models). There is going to be a long education process here - no shortcuts, but I think it will be worth it in the long run, both for them and for us (as a supplier).

  11. I see Cloud computing a part of the future. Some types of apps fit well others don't. The difference between now and the old mainframe models are the bandwidth required since MF used text. So bandwidth enables more but limits what can go in the cloud too. Second security is an issue and viturual clouds helps to separate parties, but hacking is always an issue if you want full access from everywhere. The virtual clouds also help with the political issue of control. Let IT run the cloud but make the marketing and CFO groups feel they control the structure of their clouds. So not only are their technical segregations for types of use and access, there are political issues and barriers to migrating to a cloud environment. Given this I expect smaller companies to lead the way, not the big companies, but the big ones gain more in benefits from the points listed in the blog.

  12. Very interesting topic indeed. Ms. Warrior said: The truth is, the cloud model is still nascent and most of the enterprise customers we talk to agree that service levels are not yet reliable and robust enough to be considered enterprise-class. In addition to service levels, the other two main barriers to broader cloud adoption in our view are security and the issue of interoperability and vendor lock-in. We believe the network can play a central role in addressing all three of these shortcomings in today’s commercial cloud offerings.""I could not agree more. As another one of Cisco's innumerable champions, Monique Morrow, said on her blog in recent past ( From India to Intercloud: ""Perhaps one approach is to consider a systems view especially around security itself ;and cloud computing/inter-cloud e.g., look also at:"" Absolutely spot on, same as Ms. Warrior, especially since Cisco's ( and Monique's of course) awesome experience in security and IP NGN/MPLS based services. Cisco's own vision of Cloud computing, as is reflected from the links above and the following white paper is a very dynamic vision growing out of network-is-the-computer and Cisco's IP NGN concept. However, I cannot help but wonder if there are any thoughts on issues of how ""efficient"" Cloud Computing can be, in other words: issues of data motion ? For example, in the next couple of years which country will have the resources to support Cloud computing and cloud/Web platforms technology ...? The further the distance from the ""cloud"" the larger should be the delay. The whole point of the cloud is that we are not paying for the hardware or the bandwidth so the cloud could be physically anywhere. In other words, can we beat the generic laws of Physics and solve the data motion issue? Of course, security and control over the data placed in the cloud will be a bigger concern and that has already been addressed by Ms. Warrior and her colleagues at Cisco.To summarize, my personal ( and totally unimportant !) concerns, besides the security etc. are the following:1. The simple laws of physics dictate that cloud computing will require geographic proximity to the user base. No one knows that better than Cisco. Note, this means the consumer of the app and not the provider, so countries with large web audiences and rapid growth will be key. 2. Labor costs are a minimal issue. Cloud computing isn't just a business model, but a set of technologies that replace much of the labor in data center operations. 3. Cost of Power and high-speed pipes ( of course, up goes the demand for high speed pipes up goes the business for Cisco :-))4. Legal issues, for example, the Patriot Act prevents many foreign companies from using US data centers. And some other that may not be relevant at this point in time :-)Am not sure if the above makes any sense at all. In any case, Thanks a lot for reading my thoughts if any of you did!Humbly Yourssantanu"

  13. It’s clear in my mind that the main barrier will be business sensitive data and IP being hosted off-site. It does happen (, but it will stifle growth considerably. It will be 10 years minimum before there is any movement off-site in the Banking/Finance industry. Typically these organisations are large enough or wealthy enough to host their own clouds however.Proving and promoting the security measures will be key.

  14. Main barriers for large scale enterprises to get on CC?1) Cost (because of Moore's law, this won't be an issue in the future)2) Not enough proven case studies3) Security / Privacy issues4) Ownership of data / content in the cloud?5) Spent too much on existing infrastructure and at present not justifiable yet to migrate to the cloud.Right now the drawbacks outweigh the advantages for a majority of businesses. It will take time before we see more businesses move towards CC. Once MS Office Live kicks off soon, i'm pretty sure more people will be more confident to move towards CC.

  15. All my stated worries aside, the cloud concept might offer to businesses small and large, broad use of GIS (geographical information system) services. I.E., throw your customer data sheet onto a specified geo dataset and voila a simple map, or better, a delivery system, or even better, economic profiles from census blockgroup data could be output. Load potential business locations and find sweet spots according to transport routes, pop, income and other geo data. Beats having to buy, maintain, program and then hire a GIS expert. Might level the playing field some.

  16. It is clear that cloud computing is the future - think the main issue companies have is security in the cloud. That being, feeling comfortable that the company they are dealing with will not go out of business and loose all their data or that their data will, otherwise, be lost / compromised. Once companies feel comfortable with this, the move to the cloud will be swift.

  17. This is about trust, security,availability, access speed, worries about support. This is an initial state since there are no proven record(s) how it will really work. @fcseh

  18. From my perspective, some of the key barriers to cloud computing are interface and circuit/bandwidth costs. Once the price window slides as it has done in the past, cloud computing will be more prevalent and viable to the masses.

  19. It will be useful to clarify what category of IT apps may 'not' be appropriate for the Cloud Computing approach. That will add realism and credibility to the Cloud Computing discussion.

  20. As I said earlier, the concept is definitely the next step in computing. However, there is a large number of GIS clients whose every project is composed of data streaming in from several sources, the pathnames of which cannot be changed one iota or the whole project is dead in the water. What kind of safe guards will assure us that data will not be moved around if placed in the hands of outsiders. It is difficult enough to secure data structure in-house, sometimes. Or will clouds only offer the applications and not the storage of project data.

  21. Vasanth posed a good question regarding cloud computing’s ability to “match the amount of physical resources the stand-alone enterprise provides.” Private clouds essentially create a trusted compute utility under the control of the enterprise. So IT teams will have the ability to reach out and leverage the resources of service providers in a very flexible, on-demand fashion – yet still maintain full control of a highly secure IT environment. Our Unified Computing System was designed from the ground up to enable virtualization at scale. We’ve taken many of the principles of networking -- federation, open systems, interconnect, peering, dynamic self-configuration of provisioning -- and we’ve brought those capabilities to the systems world. We end up with a system that scales and is very cost-effective – a system designed for any workload to run on any server, at any time with all of the provisioning being done in software.

  22. Several of you (including Clarke, Thomas, Simon and Aydin) make very valid points about the risks and challenges associated with using today’s public cloud offerings in enterprise environments. The truth is, the cloud model is still nascent and most of the enterprise customers we talk to agree that service levels are not yet reliable and robust enough to be considered enterprise-class. In addition to service levels, the other two main barriers to broader cloud adoption in our view are security and the issue of interoperability and vendor lock-in. We believe the network can play a central role in addressing all three of these shortcomings in today’s commercial cloud offerings.

  23. Lots of great comments here on private cloud computing! Thanks to everyone for your insights and input.Teri made an important point in her post regarding the issue of “relying on Internet functionality for personal or individual workstation operations.”There’s a lot of debate these days regarding which is the right model: on-demand vs. on-premise? At Cisco we believe that the best approach is to have a combination of different types of applications – some that are on-premise and others that are on-demand. As an industry, we will eventually move beyond this conversation, because ultimately it’s the user experience that really matters. The key is to provide an experience that’s consistent, seamless and transparent to the end user---such that we are always connected and always able to access the services and functionality we need, regardless of where we are or what device we’re using. This means that applications will be hosted and delivered through a combination of on-premise and on-demand networks (both Cisco and non-Cisco) that work seamlessly together.

  24. For the most part, I'd say 90% of services provided by IT can be cloud-ified. But high I/O services, cannot; they have to rely on their own replication solutions. Though to some degree it's relative to how you control the I/O limits to departments and/or how up to date your back end architecture is. We keep our DBs on near bleeding edge hardware; & to virtualize them would decrease the I/O performance. We do push a replica of them to the cloud; but don't rely on them other than for DR.we've been running a cloud"" for a few years; prior the buzzword existing; by replicating SANs between geographical locations: Americas/Europe/Asia."

  25. I believe there will be certain push back from IT departments, just because they will see as a thread. We saw similar situation with adoption of VoiceXML as a standard - suddenly all the IVR people were afraid they are loosing jobs since it will be now so simple to modify and develop IVR interaction.We see today at customers that certain services, that are offered by the SaaS model make perfect ROI, security and uptime are absolutely comparable with on-premise deployment but IT departments push back since they have the impression they are loosing control.Simon VostryCEO / ZOOM International

  26. I think the two priority issues with moving to the cloud are security & bandwidth constraints.Moving to a cloud model requires a complete rethinking of information classification and the ability to easily and dynamically identify what is in the cloud and how it is protected. The big issue will be the ease at which it will be possible to unintentionally leak corporate IP.We also need to remember that the majority of organizations are bandwidth constrained. One reason why companies continue to work alot on a local model is that many offices be they remote or not just don't have the capacity to use or move data across the network in a cost effective ratio. Thomas V. FischerSecurity Architect (independent)

  27. with google coming up with cloud do your percieve that computing online will become the order of the more thing i heard about is cloud computing can not match the amount of physical resources the standalone enterprise provides ..

  28. You are definitely on the right track! I've been preaching this strategy to my customers ever since VMWare started gaining ground.The next step is services in the private cloud. You start with low end services, such as disk space, database, or CPU time. Then you move to high end services - enterprise applications on demand.For example, a user requiring a CRM system may push a button and CRM will be provisioned for him in the cloud. He wouldn't worry about which data center or which server or database is being utilized. This approach will show the real face of utility computing - paying only for the services you use. If you buy cloud services from Amazon, Mosso, GoGrid and the like or if you build your own cloud, you are still talking about some fixed costs or perhaps variable costs to cover things you don't really need. With services in the cloud, you are billed for the actual use.This will work fine for public clouds but will be even more effective for private clouds.Matt ShlosbergCEO / TeraNext

  29. We at Lumison have been watching the Cisco UCS story with interest.From late last year, we embarked on a Private Virtual Cloud strategy for deploying hosted Microsoft apps (first Exchange in Feb 09, and OCS later this month), using HyperV and a Cisco switching fabric.The key issue however is to have a clear commercial / billing model. Currently, it is unclear whether the yte""-sized billing models of some cloud providers is sustainable (are any in profit yet?).Bear in mind that with the efficiency of virtualisation comes the risk of under- or over- capacity. The former risks poor service for users, the latter is not sustainable. With dedicated platforms the customer made the investment, with clouds the service provider does (certainly for ""on-demand"" services, rather than private clouds built to order). How this all looks going forward is yet to play out fully.Watch this space...!"

  30. it's a very important development, i believe it would help Cisco beat huawei easily^^

  31. Love the idea of enterprise cloud service,and cloud packaging for reorganization and expansion purposes, but cannot accept having to rely on internet functionality for personal or individual workstation operations.