Coffeehouse Thread

27 posts

The cloud taking over

Back to Forum: Coffeehouse
  • User profile image
    rjdohnert

    Do you guys think that the cloud is taking over?  Do you think its pie in the sky?  What applications do you see getting killed first?

     

    Discuss

  • User profile image
    W3bbo

    I don't see any apps being killed, there will always be demand for applications that work independently of the Internet for use in disconnected systems or enterprise systems where they don't want to outsource their data to any third-parties.

     

    We've had cloud-based email since the mid-1990s, but Microsoft Exchange Server isn't going anywhere, and GMail has been successful overall, but Google Apps haven't won over that many companies.

     

    Instead we're seeing the rise of new applications, social networking mostly, that require the always-connected paradigm and so are best suited to running in the cloud.

  • User profile image
    figuerres

    Right now a lot of folks are looking at the "Cloud" as a hype/buzz  thing and need to see if it will prove it's worth.

    while we have had Internet based services for a long time the idea of letting some other entity take more of the code and data that is vital to a business and keep it running, and safe and prove that the business model will make money .... a lot of folks want to see that work before they jump into it.

  • User profile image
    Pace

    With the whole better Mobiles / iPad craze at the moment I think things are leaning towards closed systems as such.

     

    Facebook, Apple App Store, Android Market to name but a few which have each proved highly successful.   

     

    Interesting read along these lines; 

    http://www.telegraph.co.uk/technology/internet/7954157/Ofcom-data-shows-how-technology-changes-our-web-habits.html

     

     

  • User profile image
    davewill

    I don't see the cloud taking over per-se.  The cloud is just another option.  Businesses always want to be in control of their own processes and data but there is always the need for cost savings.  I think businesses who want to get rid of their servers and high end IT staff to save costs are the ones who will look to the cloud for cost savings.  For any specific application segment to survive it needs to be able to follow the demand.  For we developers it means we need to be ready to possibly rearchitect our systems to be both on-premise and cloud capable.

  • User profile image
    spivonious

    The cloud will never take off because there is no way that businesses will trust their confidential information to be sent back and forth over the Internet. They also still want to be able to work if the Internet connection goes down.

  • User profile image
    ScottWelker

    spivonious said:

    The cloud will never take off because there is no way that businesses will trust their confidential information to be sent back and forth over the Internet. They also still want to be able to work if the Internet connection goes down.

    Not sure I'd say *never* but your point is consistent with my experience. Of course you could partition your apps with confidential data kept local but, still, it's a hard sell - in my experience. It will take some time for business leaders (technical and non-technical alike) to warm up to the idea.

  • User profile image
    brian.​shapiro

    THE CLOUD

    By Stephen King

  • User profile image
    Charles

    brian.shapiro said:

    THE CLOUD

    By Stephen King

    Smiley

     

    Another trend we'll see is that of moving computation skyward. So, one of the promises of "cloud computing" is less the cloud aspect, per se, and more on computing -> Being able to farm out complex computations/calculations over giant data sets to 100,000 virtual computers, for example, is quite appealing (and there's still a lot of work to do across the cloud computing landscape to pull this off in ways that are reliable and efficient given a number of variables...). Further, imagine being able to process large media files (multiple encodings) in minutes, not hours.

     

    The cloud will be more than a means of offloading IT functions, processes, management and server applicaitions to distributed virtual environments managed by thrid parties (of course, this is VERY appealing for a number of reasons....).

     

    The future of computing is also what's going here, not just the future of business information technology.

     

    Sky.Compute();

    C

     

     

  • User profile image
    spivonious

    Charles said:
    brian.shapiro said:
    *snip*

    Smiley

     

    Another trend we'll see is that of moving computation skyward. So, one of the promises of "cloud computing" is less the cloud aspect, per se, and more on computing -> Being able to farm out complex computations/calculations over giant data sets to 100,000 virtual computers, for example, is quite appealing (and there's still a lot of work to do across the cloud computing landscape to pull this off in ways that are reliable and efficient given a number of variables...). Further, imagine being able to process large media files (multiple encodings) in minutes, not hours.

     

    The cloud will be more than a means of offloading IT functions, processes, management and server applicaitions to distributed virtual environments managed by thrid parties (of course, this is VERY appealing for a number of reasons....).

     

    The future of computing is also what's going here, not just the future of business information technology.

     

    Sky.Compute();

    C

     

     

    No matter how powerful the cloud is, it will always be limited by the network connection. I could have the cloud convert a blu-ray ISO to an h.264 MPEG4 in seconds, but I would spend hours uploading and downloading. Remember that over 10% of the U.S. do not have access to broadband, even if they wanted it.

  • User profile image
    Charles

    spivonious said:
    Charles said:
    *snip*

    No matter how powerful the cloud is, it will always be limited by the network connection. I could have the cloud convert a blu-ray ISO to an h.264 MPEG4 in seconds, but I would spend hours uploading and downloading. Remember that over 10% of the U.S. do not have access to broadband, even if they wanted it.

    Good point! However, this will not always be the case - the physical Internet, too, will evolve and it is cloud computing that will drive this evolution Wink

     

    C

  • User profile image
    PerfectPhase

    Charles said:
    brian.shapiro said:
    *snip*

    Smiley

     

    Another trend we'll see is that of moving computation skyward. So, one of the promises of "cloud computing" is less the cloud aspect, per se, and more on computing -> Being able to farm out complex computations/calculations over giant data sets to 100,000 virtual computers, for example, is quite appealing (and there's still a lot of work to do across the cloud computing landscape to pull this off in ways that are reliable and efficient given a number of variables...). Further, imagine being able to process large media files (multiple encodings) in minutes, not hours.

     

    The cloud will be more than a means of offloading IT functions, processes, management and server applicaitions to distributed virtual environments managed by thrid parties (of course, this is VERY appealing for a number of reasons....).

     

    The future of computing is also what's going here, not just the future of business information technology.

     

    Sky.Compute();

    C

     

     

     I've heard a lot of people say you'll just be able to spin up 100,000 instances compute your problem and release the instances. 

    Wondering how much standby capacity these data centers will have.  If you only have customers adding and releasing a few instances at a time then things balance out, but if you have a small set of users causing huge peaks then there must be a large amount of standby capacity not earning money for the hoster?

  • User profile image
    ryanb

    Charles said:
    spivonious said:
    *snip*

    Good point! However, this will not always be the case - the physical Internet, too, will evolve and it is cloud computing that will drive this evolution Wink

     

    C

    Cloud computation does have a lot of interesting potential, but as for cloud data management, it only applies to a certain set of problems.  Cloud computation would be a great tool for things like the SETI project, decoding the genome, worldwide access to huge data sets like environmental change data analysis or processing data from the LHC, etc.  Proprietary research facilities, however, are going to have the same sort of security concerns to deal with, and often have adequate computing resources available to them already.

     

    Cloud data storage/management is a great resource for social platforms (Facebook), open-source programs, etc.  I see the primary business customers as the small to medium companies that don't have the resources for full-fledged IT departments managing servers and related systems.  For the reasons mentioned, very few large businesses are going to be convinced to hand over their data to such systems, and rightly so.  Yet large business is always where the real profit potential lies.

     

    I think that if MS (or anyone) wants to find success with cloud computing in large business, they need to target efforts into business-private cloud systems.  Companies are looking for reliable, secure, easy-to manage solutions that enable working within their own company (often including branches around the world), but with the control and security of keeping everything on company owned and managed servers.  The supplier that provides the best private-cloud solutions to large businesses is the one that will win.  Targeting joe-homeowner users (Google) is fine but isn't going to keep the lights on and pay the investors.

     

    There is a lot of untapped potential in cloud stuff, and I will be interested to see where it goes -- as long as it is done in a realistic/relevant way.  However, there is absolutely zero chance of it "taking over" or removing the need for local computing.  Any such ideas are silly dreams that have been repeated and proven wrong for decades now.

  • User profile image
    Charles

    PerfectPhase said:
    Charles said:
    *snip*

     I've heard a lot of people say you'll just be able to spin up 100,000 instances compute your problem and release the instances. 

    Wondering how much standby capacity these data centers will have.  If you only have customers adding and releasing a few instances at a time then things balance out, but if you have a small set of users causing huge peaks then there must be a large amount of standby capacity not earning money for the hoster?

    The standby capacity is a virtual problem: the "computers" that spin up are VMs.

    C

  • User profile image
    elmer

    Charles said:
    PerfectPhase said:
    *snip*

    The standby capacity is a virtual problem: the "computers" that spin up are VMs.

    C

    Where I work, Reliability and Data-Security remain our big concerns, and two recent events reconfirmed our reluctance to commit anything "mission critical" or to hand over control of our data.

     

    1. Both redundant ISP links were taken out by a major failure at our local exchange, and we lost all internet service for the best part of 2 days. Email and Web were totally off-line, but at least people could still work.

     

    2. During a "maintenance" task on our hosted VPS, the provider enabled an unprotected public link without any referal to us. I just happened to be checking things over and discovered a heap of remote connections to port 135, and then traced the cause. Fortunately there was nothing private on there and we were lucky, but nevertheless, I'm not happy.

     

  • User profile image
    PerfectPhase

    Charles said:
    PerfectPhase said:
    *snip*

    The standby capacity is a virtual problem: the "computers" that spin up are VMs.

    C

    No, the standby capacity is a physical problem.   VM's still need physical compute nodes.  If your going to run up 100,000 virtual nodes, asumming each has 2 virtual cpu's on quad processor quad core physical machines, that still means you need need 12,500 physical machines on standby.  It could be less if you overcommit the physical machines, but I don't think Azure does and it would kind of suck for high CPU tasks.

     

    I'm saying if you'r selling the promise that you can grab a lot of machines on demand, then you'll have a much higher delta between peek machine use and the low points.  And while the physical host machines are sitting unused, they are costing the hoster money and not making any.

     

    What I'm really questioning, is if we'll ever get to the point that you can at zero notice request 100,000 compute nodes and reilably expect to get 100,000 or if you'll have to book that size request in advance.

  • User profile image
    PerfectPhase

    ryanb said:
    Charles said:
    *snip*

    Cloud computation does have a lot of interesting potential, but as for cloud data management, it only applies to a certain set of problems.  Cloud computation would be a great tool for things like the SETI project, decoding the genome, worldwide access to huge data sets like environmental change data analysis or processing data from the LHC, etc.  Proprietary research facilities, however, are going to have the same sort of security concerns to deal with, and often have adequate computing resources available to them already.

     

    Cloud data storage/management is a great resource for social platforms (Facebook), open-source programs, etc.  I see the primary business customers as the small to medium companies that don't have the resources for full-fledged IT departments managing servers and related systems.  For the reasons mentioned, very few large businesses are going to be convinced to hand over their data to such systems, and rightly so.  Yet large business is always where the real profit potential lies.

     

    I think that if MS (or anyone) wants to find success with cloud computing in large business, they need to target efforts into business-private cloud systems.  Companies are looking for reliable, secure, easy-to manage solutions that enable working within their own company (often including branches around the world), but with the control and security of keeping everything on company owned and managed servers.  The supplier that provides the best private-cloud solutions to large businesses is the one that will win.  Targeting joe-homeowner users (Google) is fine but isn't going to keep the lights on and pay the investors.

     

    There is a lot of untapped potential in cloud stuff, and I will be interested to see where it goes -- as long as it is done in a realistic/relevant way.  However, there is absolutely zero chance of it "taking over" or removing the need for local computing.  Any such ideas are silly dreams that have been repeated and proven wrong for decades now.

    Putting data saftey to one side.   I can see cloud compute being quite hand for people like SETI and the LHC if used with their in house data center.   Those people will have a permanent need for compute cylcles and it makes more sense for them to have enough in house horse power to meet their base line compute cycles, it will be cheaper at that scale than a cloud provider, but the peaks should be out sourced on demand to the cloud.  For example if the LHC had a really intreasting days run, they may want to expidite the processing of the data set, then it makes sense to call down another 100,000 compute nodes to process that batch in a hybrid cloud configuration.

  • User profile image
    Charles

    PerfectPhase said:
    Charles said:
    *snip*

    No, the standby capacity is a physical problem.   VM's still need physical compute nodes.  If your going to run up 100,000 virtual nodes, asumming each has 2 virtual cpu's on quad processor quad core physical machines, that still means you need need 12,500 physical machines on standby.  It could be less if you overcommit the physical machines, but I don't think Azure does and it would kind of suck for high CPU tasks.

     

    I'm saying if you'r selling the promise that you can grab a lot of machines on demand, then you'll have a much higher delta between peek machine use and the low points.  And while the physical host machines are sitting unused, they are costing the hoster money and not making any.

     

    What I'm really questioning, is if we'll ever get to the point that you can at zero notice request 100,000 compute nodes and reilably expect to get 100,000 or if you'll have to book that size request in advance.

    Good questions. I'd be disappointed if it doesn't someday become the case. I mean, we are so early in the cloud computing game that one has to assume technological advancements on the order to accomplish what we're talking about here -> "Please process this on 100,000 CPUs. Thank you." Smiley

    C

Comments closed

Comments have been closed since this content was published more than 30 days ago, but if you'd like to continue the conversation, please create a new thread in our Forums, or Contact Us and let us know.