What is The Cloud?

In this presentation we're going to attempt to answer what has become a very very popular question in the last few years. What is this cloud that everyone is talking about and using and why is it such a big deal. But before we jump to present day let's actually hop in a time machine and go back 100 years ago before the invention of the electric grid. A lot of individuals draw parallels between the electric grid and the innovation that it sparked and cloud computing and the innovation that it is sparking currently. A hundred years ago if I wanted to build widgets or power machines in my factory I needed to build my own electric power station. If we zoom in here we see a picture of a dynamo from this era. And to give you a sense of scale this is a person over here on the lower left. Now these dynamos were incredibly expensive to both create and maintain their parts came from a limited number of suppliers and those suppliers could pretty much charge whatever they wanted to both build and maintain that power station. It doesn't take a huge leap of the imagination to draw parallels between this dynamo and your data center the computing resources that you consume in your data center come from a limited number of suppliers are very costly to build and maintain and there's always a chance that your system administrators could put things together the wrong way and break everything. In his book The Big Switch Nick Carr called out exactly this relationship. He says that cloud computing today is what the electric grid was 100 years ago and anyone who wants to build little cloud computing should really take a close look at how the electric grid changed the face of the world. When companies could start plugging into electric power to drive things like blenders TVs mobile devices so on and so forth. The world changed overnight and it sparked 100 years of innovation the likes of which the world has never seen. A lot of folks think that cloud computing is going to change the face of the planet the same way when we can plug in to computing resources the same way that we plug into a socket in our wall. Now it's jumped back to present day and actually define the cloud. I've looked high and low but I've never found a definition that really suits me. And no it may be arrogant. I've come up with my own. I think the cloud is really using virtualization and automation to provide users with self-service computing products that provide infinite's scale at low cost. That's a really long sentence with a lot of moving parts. But if we break it down into just two pieces the cloud is really virtualization and automation. Let's take a second and zoom in on each of these virtual just means not physical. So C-p you disk RAM and network cards are virtual sized so that they can be used by many virtual computers on that machine. Let's zoom in over here on the left and take a look at exactly what we're talking about down here in the very bottom. We have an actual physical machine. It has a limited number of C-p use memory network cards and disk. When we talk about virtualization what we're actually doing is plugging in this hypervisor here in the middle so that we can chop this C.P. you ram and other resources into individual small little pieces that many consumers can get ahold of. Here we're saying that we created 3 CMs or virtual machines on top of this one single physical machine to each of the users who are using these CMs. These appear as physical machines I can log into the server. I see a smaller subset of those resources and I have access directly to each of these resources. This is where virtualization is. Software can also be virtualizing so that I don't need to have my own copy of it and many users can share it. Virtual hardware and software is incredibly easy to spin up and take down much much easier than physical. A lot of folks are going to take a look at this and say Hold on we've been virtualizing for a long long time. What does all the hype. Why is this such a big deal. Well a lot of things have changed recently. Hardware virtualization just started becoming possible in the late 1990s. The broadband and mobile space have been driving what we call Web scale where not only do we have maybe hundreds or thousands of people accessing our services at the same time we may have hundreds of thousands of people the hardware speeds and capacity have only recently changed to allow US companies like Amazon and Rackspace Cloud just started offering their services recently. There's also been this whole SAS or software as a service crowd with companies like sales force and Google apps coming out of nowhere and really upsetting the marketplace. You also cannot underestimate the economy these days folks are trying to do a lot more with a lot less and the cloud enables that. On the automation side automation gives us self-service and the illusion of near infinite scale where not only can I spin up a single server but I can spin up a thousand servers in that same amount of time. Automation also gives us access to utility based pricing and service. If we draw parallel here I can go through every room in my house and turn on all the lights if I'm having a big party or turn on a light in a single room if I'm reading a book in the cloud space I can go on and turn on 100 servers when I'm really getting hit hard and take that capacity down to a single server when I'm not to take a look at the self-service kiosk model. Let's zoom in on Amazon's console here you can see that it can spin up anything from servers to DNS to databases and storage and so on just in a couple of mouse clicks. No longer do I need to go in and procure a server wait for that server to arrive set the server up and then install it in Iraq I can get access to the same end product in just a few mouse clicks. Any discussion of what the cloud is has to include the three different cloud types public private and hybrid public is what services like Amazon offer low cost least virtual servers. It gives us the illusion of near-infinite scale and gives us access to that utility based pricing we were just talking about this is public cloud private cloud is where I grab hardware I pull it in-house I set it up and I set up my own cloud for my own internal users. Now even though they can spin up servers on demand I do have a limited scale and I no longer have that infinite scale that we were talking about. Those servers are going to come with a fixed monthly cost. I'm going to have to purchase bandwidth. I'm going to have to call them and I'm going to have to replace parts when they break. On the flip side I get total control with the private cloud with a hybrid cloud. I leverage both public cloud resources perhaps to burst capacity and I'm able to still leverage my internal fixed cost and total control private cloud. If we look at a Venn diagram over here on the right you can see that we have public cloud and private cloud hybrid kind of sits in the middle between the two. We also have a concept of a cloud pyramid where we have foundational services on the bottom and more higher level esoteric services on the top. These are typically grouped into infrastructures a service or I guess on the bottom platform as a service or P-8s in the middle. And software as a service or as a asset at the top in this course we're going to look exclusively at infrastructure as a service or I guess this is essentially raw computing power or raw computing services things like databases queues email servers and raw servers themselves. Amazon's products such as easy to Rackspace Cloud go grid joint and others are companies who play heavily in the infrastructure as a service space. Amazon themselves is arguably the 800 pound gorilla in this space. Amazon groups about 30 independent services into what's known as Amazon Web Services. These are many different computing services that are subgroups into compute storage networking and so on. Probably one of their more popular services is elastic compute cloud or easy. These are those virtual servers we've just discussed starting at about $600 a month. All of their services are on demand with self-service startup and shut down and many of these services also feature something called auto scaling which automatically let those servers and services preve up to a higher capacity when a higher demand comes in to themselves. Also include our These snapshots. So we have point in time rollbacks of what our server looked like at any given time. And it also makes very very cool things like h a high availability and high volume. Extremely simple to use in some cases as simple as a checkbox. If we look at how many smaller businesses deploy their services we typically see something like this. It's cost prohibitive to buy a whole bunch of physical machines so they have a single machine serving as their everything. It's their web server their database perhaps their email server their quarter Pozza 3:15 and everything else. Each of these independent software packages have specialized security or speed needs but because they're all housed in one single server we can't optimize any of these layers. Ideally we'd like to see something like this for a web application where we have not only independent rails of services. For example here's a left rail and right rail we also have tiers. We have a load balancer on one tier a web server an application server and another tier and a database in a third tier. This enables high availability meaning that we could lose one entire rail or one independent machine and still be able to serve our traffic and it also enables scalability. For example if our web server or application server started getting overloaded we could scale up more servers at this level independently of our load balancer in our database. Again the reason that a lot of businesses don't do this is because it's cost prohibitive. The cost in the physical world to purchase these servers would on us at least $25000. The cost in the cloud is a pay as you go at about 150 a month. Hopefully that quick tour of these various areas give you a good overall definition of what the cloud is. Now we can dig deeper in the other presentations.