Cloud computing isn't much different from time-sharing on mainframes, for those of us who are old enough to remember those days.
An interview with Rob Wray and Angelo Rajadurai from Sun.
Robert Wray: Hi, my name’s Rob with MP3Car, and we’re here with Angelo from Sun. Angelo, maybe you can tell us a little bit about what you do at Sun before we start talking.
Angelo Rajadurai: Hi Rob thanks for the opportunity to talk, and I’m pretty excited to spend a few minutes with you guys. I work in the Sun Startup Essentials program. I’m responsible for the East Coast implementation of this program. I also work in the Sun cloud team, and it’s very important for Sun that startups actually adopt the cloud. And we’re seeing a lot of startups actually working on the cloud, and we want to make sure that startups can use and benefit from Sun’s cloud effort as well. That’s in a nutshell what I do at Sun.
Robert Wray: Well thanks a lot for taking the time to tell us what you’re doing and talk about Sun’s efforts. So in previous blog posts we’ve talked about some of the things that we think are going to happen in mobile computing over the next five years, so what role could cloud computing play in solving these demands over the next five years?
Angelo Rajadurai: So interestingly enough what we’re seeing is, you know the cloud is maturing really, really quickly and especially in the space of startups where you see all these great innovations happening. Cloud is making life much easier for new companies to come in and provide services, so especially in the mobile space where it’s again a budding technology where there’s a lot of these services that are possible.
And in the fast days it was very hot to actually provide these services because it took you, you know you had to buy a whole bunch of hardware, you had to setup data centers to be able to provide the service, but with a cloud now what we can do is you can provide these services in a much cheaper cost and have multiple of these services kind of come up really quick. So it’s actually quickening the pace of innovation in the space as well, so there’s lots of these services that are coming up, and I think cloud can really help especially in the mobile space where your device on your car is just an access point to big servers that can be hosted on a cloud. And a cloud provides you all this great compute power that you can use to deliver such a service to a device like a device on a car.
Robert Wray: Well what are the biggest challenges you see with developers using cloud computing to support some of these advanced features that people are going to demand on the go?
Angelo Rajadurai: So right now what I am seeing is there is no set standard for actually exposing these services to the access points. So that problem is kind of getting solved on the compute side a little bit because there are all these standards that have emerged and people can plug-n-play and mash these contents together if you may. So for example you could add today on your desktop a Google map, which is a service, with maybe a real estate listing service or something like that, and you could create these mashables really quick because the standards are already well established, and the whole concept of plug-n-play works very well.
I don’t see that happening yet in the mobile space, so I think you need to custom put these services together. Typically you’re gonna see a lot of these services that show up, but I think the value is in adding and aggregating these services and kind of providing a super service if you may. And for you to do that I think you need to establish standards, and I think that’s one of the areas where things are lagging a little bit.
Robert Wray: So what’s Sun doing to drive standards and help developers utilize the cloud as it relates to mobile computing or consumer services specifically?
Angelo Rajadurai: So if you look back at Sun’s history one of the things that we’ve been very good about doing is to kind of define and drive standards, and Sun from its inception kind of believed that the computing is more than what you can actually accommodate on your desktop or on your servers. So we, by default, put in these network access points into every machine we sold. So for the cloud computing as well, what Sun is doing is Sun is actually defining the standards for the cloud so that it will allow you to kind of seamlessly move between what you’re calling a private cloud into the public cloud.
So you could average in a service that runs locally on your own cloud or your own hardware that you have on your car if you may, and then when you are connected to the cloud, you can seamlessly migrate it to the cloud and accommodate services from there. We’re not doing this specifically for the mobile world, but this problem exists not just in the mobile space, but also in the traditional computing space where people are kind of finding that the cloud is great for certain purposes, but for certain purposes it makes more sense for us to create our own private clouds. And we need to be able to seamlessly move between these two clouds, and Sun is now defining the standards. What we’ve done so far, we’ve published a standard for this and it’s in the publish phase, and we’re gonna have our implementation of the standard come out in a few days, which will make it a little bit more interesting in the space where standards are concerned.
Robert Wray: So how do you think clouds are going to contribute to a customizable mobile experience? So what I mean by customizable mobile experience is based on what you’re doing, if you are walking or if you’re taking the car, or you’re sitting down you’re gonna wanna different type of user experience. Also different users are gonna want different types of brandings, different colors, other things like that. What role do you think clouds will play with these customizable user experiences?
Angelo Rajadurai: So from a point of view of the cloud – so you can define the – and at least in my mind and am no expert in the big mobile space, but in my mind I see that the customization happens at the access point. So let’s say you have a device that you put in your car, that access point actually defines your point of customization. What the cloud can do is because of the vast compute resources that you have at your disposal you can actually provide information from your access point.
Let’s say – tell the cloud service that you are now moving at 80 miles an hour or something like that, and the cloud service can at that point decide what kind of content it wants to provide you. Obviously you don’t want it to be streaming video to you at that point because it makes no sense to do it, but if the cloud service knows that you’re sitting with the same device sitting on your - you know in a rest area and not moving, you could potentially send a service for video maybe to your device. So I’m just coming up with this example here, but the idea here is that because the cloud can vastly configure itself and change on the fly you’re not limited to one or two different services that you can fit into your access point.
So even though the cloud - I don’t see cloud kind of doing the customization because of the vastness of the resource. You can customize based on what your access point is telling the cloud. I hope I’m making sense here, but the thing is now you can think beyond your little device, and the cloud is a vast place where you can particularly have a lot of these services available for you. And you can decide which one you want to plug in to based on some configuration that either the user has set or that automatically goes to the cloud for it.
Robert Wray: Right, okay. Yeah, it makes a lot of sense. So a lot of people, just to get back to what Sun’s role is in cloud computing and consumer stuff, a lot of people think of Sun and cloud computing architecture as something that’s just available in the enterprise; what open consumer oriented cloud models could developers look towards for inspiration?
Angelo Rajadurai: So Sun has traditionally been viewed as an enterprise, but if you go back even a few years Sun started off actually working with startups and smaller companies and consumers, and only in the last ten years or so where our focus has completely turned into enterprise. And we are trying really hard to get back into the startup space with the acquisitions like MySQL, which is very predominant in the small and medium sized businesses as well as in the personal space if you may. And with different things Sun has done in the last few months or maybe last couple years, we’ve really kind of focusing back into the end-user space if you may. And as far as the cloud is concerned the income in today’s - basically Amazon, right, it’s a great service, they kind of defined the space where it’s really open to the end-user as well.
Sun’s cloud when we announce it would be open to both end-users and enterprises as well, but that line is kind of merging really quickly. In maybe five or six years ago, you needed to be an enterprise to work with Sun because the cost of hardware was so high, and you had only one option that you could buy it. There was nothing else that you could do, but right now with the cloud that line is merging. It’s very cheap for an individual to kind of get together a 100 machines in the cloud and provide a service at $.10 an hour or so, right? So that line between enterprise and individual is quickly kind of going away and the cloud is equalizing that space as well. So what Sun can do is it can bring the expertise that it has developed for enterprises and building these large server-farms and make it accessible to end-users. So I think Sun’s exposure into the enterprise space is actually going to be a beneficial one while we bring the service to mobile users.
Robert Wray: Are there any specific examples where - we talked before about standards being one of the really hard problems - where there’s been lots of developers that have worked to set standards and have made consumer products or is that still yet to happen?
Angelo Rajadurai: So there are many places in the compute space where there was all these standards were set by end-users. In many cases these standards get set because one company is driving it, one large enterprise drives it and then everybody comes in. For example, if you look at the Wal-Mart case when our FID was being pushed. Wal-Mart kind of came and said, “This is the standard that you will follow for our FID,” and that became the standard. Or if you look at the ebXML space where it’s XML for businesses, the standards were set because some large company kind of blow it. But you also see a lot more examples of these, like for example in the SQL space there’s more startups like MySQL, which was a very small startup a few years ago kind of set the standard for this open source database space. So the standards are being set either by a large company driving it or by a grass root effort, which see a need and kind of set these standards.
And without setting the standards it’s really hard to build up on a really expandable space. So if you look at any of the open source projects that are successful, the first thing that they need to do is setup some kind of a standard that people can contribute into, and this has been happening for a while now. So you can see that model; in one case it happens because of some company driving it and being the income, and they get to say exactly what the want to sell, but it also happens in the open source space where it’s being set by communities, and being followed so that the contribution can be made simpler.
Robert Wray: Well other than a lack of standards, what mistakes can be learned from other failed or partially successful cloud efforts?
Angelo Rajadurai: So if you look at the cloud effort, one of the things that you really need is an ecosystem around the cloud. So none of these servers in the cloud can live on its own. So for example you can have a very successful cloud with really expandable things, but if you don’t have the services that are around it then these cloud efforts kind of fail. One of the things that you can see with Amazon cloud, which is again the standard today, is that there’s hundreds of partners and maybe even thousands of people that are providing services in the cloud.
So if you take that same example into the mobile space, any service that you’re providing needs to have an ecosystem around it. It needs to have the support structure around it and without the support structure things don’t succeed. You can have a great service, but if you don’t have like the database support around it or if you don’t have, in your case, like the location services around it or even like the developer help that is around it then those things don’t work too well. Again if you look at open source one of the things that you get by open sourcing your software is that you build ecosystem around your software, which includes people that know the software, which includes people that can help you fix and install, scale your software and so on. So that’s another important piece of any effort that goes in. For cloud and for everything else as well I think it’s pretty important that you have an ecosystem around it for services and support for you.
Robert Wray: Well looks like there’s a lot of exciting opportunities for developers that are looking to do bleeding edge stuff in the mobile computing space. So if you want to check out our blog for links, where to get started, we’re gonna have a bunch of stuff there and also transcripts of the interview. So Angelo thank you so much for your time. We appreciate it.
Angelo Rajadurai: You’re welcome Rob.
[End of Audio]
This slide is fresh from telematics update a few hours ago.
How, exactly, does Gartner say that "cloud computing" is substantially different from old-school mainframe time-sharing?
With time-sharing, you use a terminal of sorts to interface with the mainframe that delivers applications to the terminal.
Replace "mainframe" with "cloud" (or "server cluster", if you prefer) and "terminal" with "web browser", and you have the next big thing from Sun Microsystems.
I agree that industry buzzwords get diluted when they are ubiquitously and erroneously applied and the meaning is lost in the translation. However, this is not one of those situations.
I'm not saying that cloud computing is bad or couldn't apply to this hobby or anything even remotely similar.
I'm just saying that the concept has been around since the days of punch-cards and reel-to-reel magnetic tape.
I guess I'm at a loss to see what cloud computing could really do for the mobile market anyway... other than maybe lower the power and cost of some type of integrated unit by turning a CarPC into a cloud-connected terminal.
Are there any real-world examples that could show me wtf the cloud would be used for? And some type of use that isn't readily available via regular internet connection?
The best resurrected frontend I've ever used, period.
Yah you are right. At the crux of the cloud is a timesharing system. But saying mainframes and clouds are the same is like saying email and the pony express are the same. They both provide compute to multiple people from a centralized resource but they are fundamentally different things!
Sun's moto for over 30 years has been, "Network is the computer". The emphasis here is on the "network". In the last 30 years, we have been a sea of change from the mainframe era. For one, cost of the computers have dropped dramatically, network bandwidth has sky rocketed, standards have been established and innovation happens everywhere and not just at a few large corporations.
What cloud provides is a very cheap way to get access to huge compute resources. Ideas like pay per use, "realtime provisioning" are much more improved in the cloud. The cloud is more like "Server on demand" rather than a way to run some batch jobs on demand (which is what your mainframe was).
The main innovation is not "sharing compute resources" but the main innovation is the ability to create a server (even though virtual) on demand at an affordable price.
Its not just Sun talking about the cloud - there are a huge number of vendors that are providing cloud services. The cost savings of the cloud have made it an indispensable tool for many companies. Think of this. You can now pay a few dollars a month while you prototype your solution and scale it up on a need basis as your solution user base grows. No need to "pre-buy" system for anticipated needs. No need to set up your own data center. No need to worry about your server going down. No need to worry about backup etc. Its all someone else's problem. You just use it and pay for it on an hourly basis.
I think we would do well in heading the wise king Solomon's saying, "There is no new thing under the sun. (Bible - Ecc 1:9)" What is new is a form of what we have had before only much more accessible and cost effective. This is brought in part because of "standards and an open source ecosystem". This is the gist of my interview.
Hope this makes sense!
Love to hear your feedback!
Consider the capabilities that are emerging.
-Declining wireless bandwidth costs will make data connectivity and streaming very inexpensive over the next few years.
-Increasing coverage and reliability of 3G, 4G, WiMax networks will continue to improve reliability and speed.
-Mobile devices are increasing in power, decreasing in size, and increasing in storage capacity
-Web app technologies like AIR or Flash are reducing platform dependence for applications
-For the first time, inexpensive GPS allows location awareness
Converging those factors hints at a future that changes the current mobile experience from a device/software combination to a location/situation experience.
In addition, connectivity to the web allows access to immense amounts of computing power far beyond what can ever be crammed into a device. The cloud's contribution to this is to allow users to access services and data in a way that can be customized for the user.
In effect, rather than customizing the user interface as we are used to do when skinning a front end, these capabilities hint at the possibility of actually customizing the user's experience for each and every user.
The convergence of hardware and software offer a most compelling possibility to mix and match cloud services in a way that can be done by mere mortals. Don't think about those boring enterprise cloud apps. Think about how those capabilities can be harnessed for our purposes.
For example, combining a cloud service that can access my music library on my home media server with a speech to text decoder service that can interpret what I say into my phone, strip out the car noise, interpret a command, and stream the music I want to hear.
An auto-Facebook/Twitter service that could read me the tweets I'm following while posting either auto tweets based on location or voice input would be cool.
The ability to interact with those identical services via voice if in the car, and via GUI if on foot, with the cloud making the choice based on my speed and location (over 8 mph and more than 1/10th of a mile from home, use voice, under 8 mph and 1/10, use GUI). Entering the house? Switch to my home media remote control.
Why port your plug-ins from RR to Centrafuse to Streedeck to [insert FE here] when you can just write it once and make it accessible to most anyone on most any platform?
The most interesting part to me is exploring whether or not this stuff is possible. I think this community can do that better than anywhere else. Who knows more than we do about mobile needs? Yet, I often find mobile computing to be a technically difficult, often unsatisfying compromise. Is cloud computing the solution to that problem? I don't know. But it does offer possibilities worth exploring.
I can see why this might be the trend to want to go this route but this reliance on networks and someone else can sometimes be annoying too when we are worried about convenience but maybe not necessarily reliability.No need to worry about backup etc. Its all someone else's problem. You just use it and pay for it on an hourly basis.
Example, my company has a network system, when it starts to act up everyones computers crap out, NO ONE can get anything done everyone *****es and moans, productivity goes down and I'm sure IT phone starts ringing off the hook (because they can't send help desk tickets). I unplug my computer from the network and continue my work locally, plug back in when I am ready hoping to server is functioning.
Sure it's someone else's problem to FIX or make work, but it is still the users problem to have reliable access to it and the user is not very understanding when it doesn't work in this "on demand got to have it now because I am impatient" world we live in.
Everything is amazing, Nobody is Happy
Let's say I have no music stored locally on my carpc and I connect to the home network. Something happens to my connection either on the server side or somewhere in between. Guess what I have no music now, stuck on a long 24 drive to florida with my family in the car. Boy that would suck.
I don't think the need to be self sufficient will ever go away, I'm still of the mindset you can't always rely on someone else to do your work. Always have a backup plan. Not trying to be bah humbug, but nothing wrong with consulting services but definitely make sure there is a local backup plan in place because not everything works 100%.