Introduction to IT Infrastructure

[00:00:00] Speaker 1: Hello, my name is Germán Retana from Incaid Business School in Costa Rica. This video is meant to teach business executives the very, very basic things of IT infrastructure they should know. As economies become more digital, it is necessary for executives to have at least a basic understanding of how software applications work. In this video, we will cover, first off, the basic elements of an IT infrastructure, some basic vocabulary. We will talk about what are data centers, what are servers and clients, and also what network devices are. Then we're going to see how we interconnect all these devices, we're going to talk about communication media, and then we're going to offer a brief categorization of the different types of data networks we find out there in the wild. Let's start talking about the data center. A data center, like the one I show you on the right, is to start off a physical space where the servers and network devices are all located. It is usually a special place that is conditioned for this purpose, and it will have some access control for security, such that only the people allowed to touch the servers will actually do so. The servers and network devices emit heat, so we want to cool them, and there will be very powerful cooling devices in them. We also want to supply power, but not just any power, but actually redundant power supplies, and so forth. So it's a room-conditioned, prepared to host our servers, because after all, this is where our production applications and our firms are going to be running. The data centers are usually organized in racks, which can be thought of as cabinets, vertical cabinets, where we can assemble or organize our equipment. When we talk about devices that are installed in racks, we're going to be talking about rack-mounted devices. And something very important is that data centers are going to vary significantly in size. Whereas a small and medium enterprise might have a very, very small room that functions as a data center, larger firms such as Fortune 500s or technology giants such as Googles, Facebooks, Amazons of the world, will have data centers that span several soccer fields in space. I encourage you to click on the link on the bottom right of the screen to watch a video that takes you inside a Google data center. You will see that the cooling devices are not the traditional air conditioning systems we might see at homes, but actually even water-cooled systems for these tremendous data centers. The master of data center is the person of a system administrator. System administrator can traditionally be seen as this know-it-all person who really knows how every single element inside the data center is connected. This very powerful, sophisticated engineer. However, not all data centers look like the ones I've shown you on the pictures. Sometimes our data center might be a complete mess, where we don't know where things are connected, and in these cases the system administrator cannot do much more than pray that nothing falls, that nothing breaks, and that the systems continue operating. And you might think that nowadays organizing a data center is fairly simple. All what we need is a diagram. Well, let me show you how diagrams looked back in the day. This figure right here is the network diagram of Purdue University in Indiana in 1994. It's fairly complex. It's a fairly small installation. It's a university 20-plus years ago. And you might think, well, that's all we could do when we did not have computers to draw diagrams for us. So maybe our expectation is a diagram like the one I'm showing you now. This could have been made with PowerPoint or Visio, and it clearly shows the different devices, such as servers, clients, switches, access points, and also the access to the internet. However, this is a static diagram that is not too useful to system administrators. System administrators need to monitor the performance of every single node connected to a network. For this, they might use an open source tool called Nahios, which shows in green and red shades the performances of the different devices or nodes on the network. Networks are built off a series of elements that we're going to be talking about next. This diagram shows you the typical elements we're going to find in any LAN, any local AR network. And as such, they represent the devices that we might find in every single firm. If a business is using IT, chances are all of these elements, or at least most of them, are going to be present as part of their IT infrastructure. Within this network, we will find servers, we will find clients, we will find network devices, and so forth. And let's talk about each of these, starting with the ones on the left, the servers. A server, we must understand, is nothing more than a computer that serves or supplies the data and applications used by clients. Sometimes, we're going to give servers names based on the functions they perform. And in line with this, we might talk about a file server, a server that has as function to share files with potential users. Linux-based Samba servers, or more traditional file transfer protocols, FTP protocol servers, would allow us to share files with our users. These, by the way, have been replaced by modern cloud-based systems, such as Google Drive or Dropbox. We may also have a print server. This is a print server that manages the queue of printing requests by all users in a firm. You don't print directly to a printer, but rather you send it to the server, and then the server distributes the workloads through the printers in the firm. We may also have web servers. If you ever wondered, how is it that web pages are displayed when you open your browsers and your laptops? Well, the browsers are being connected to web servers. Several software can perform the function of a web server. We have open source Apache, which is by far the most popular web server in the world. Nginx is coming out as a new alternative to Apache, which has a characteristic that is very slimmed down. It performs very few functions, but in a very efficient manner. Moving to more classic web servers, we have Microsoft Internet Information Services server. We also have application servers. These are the servers that run the enterprise applications, such as an ERP system, for example, SAP, or any other application developed locally by the firm. Mail servers, which were widely used before the emergence of Office 365 or Google's Gmail, such as Microsoft Exchange or Zimbra, were used in firms to handle all the mail of the users. Another type of server present in almost any firm is going to be a database server. These servers organize the data used by all the information systems in a firm. Examples of these are Microsoft SQL Server, Oracle, or MySQL or Maria database, which are open source alternatives. Sometimes we have servers that are used for video streaming or to share photographs, to host a photo gallery, for example. These are called media servers. And we may also find servers that enable users to collaborate with each other, to work in concurrent forms such as Microsoft SharePoint or IBM Lotus. Another way we can name servers is based on their platform. And by platform, I mean their hardware and operating system. We can refer to their hardware based on the make and model. Similar to how we talk about a Toyota Corolla, the make and model of a car, we can talk about an IBM P-series server. We will also refer to the server's operating system. Just as how your laptops run Windows or Mac OS, a server can run an operating system such as Microsoft Windows Server. We could also have a distribution of Linux such as Red Hat Linux, Debian, Ubuntu, or others out there. More classic servers may run Unix as their operating systems. So putting all this together, we might find examples such as an IBM P-series with Linux, a Dell PowerEdge server running Windows. Or we could simply refer to a Linux server or a Linux to refer to a certain box, a certain server running the Linux operating system. Another way we can categorize or classify servers is based on their features or organization. And here I'm going to be talking about mainframes, high availability servers, clustered servers, or virtual servers. Let's start with the first one, mainframes. A mainframe is a very, very large multifunctional equipment. These servers are capable of running a huge number of transactions and supporting the workloads of thousands of users. They will generally cost millions of dollars and will generally be found only in Fortune 500s or big financial firms. Firms that have enough workloads to justify the investment in one of these humongous servers. If you want to learn more about these, I encourage you to click on the link that takes you on a brief tour of what the IBM Z Enterprise system technology is, which is basically how modern mainframes look like. Worth mentioning, if we were to draw a mainframe in a network diagram, the icon on the bottom right part of the screen is the one we would use to represent the mainframe in our network diagram. Then, not as large as mainframes, but geared or designed to host a firm's production applications, we have what we call high availability servers. These servers are very powerful PCs or very powerful servers that have elements that make them highly available. And what do we mean by this? For example, one of the typical things that will fail on a server is one of its hard drives. But a high availability server will have multiple of these hard drives, such that if one of its hard drives fails, the others continue performing the same functions. Sometimes these are arranged in what we call a RAID array, a redundant array of inexpensive disks, which offer redundancy to the hard drives. We will also not have a single power supply. A power supply is the element by which electricity flows into the server. We're not going to have one, we're going to have multiple of them in a redundant manner. Similarly, network interfaces or network cards can burn out, and we're going to have multiple of them in case one of them fails. The cost of these is not going to be millions of dollars, but yes, the few thousand dollars. Now, not referring so much to a single server, but rather to a group of servers, we can have what we call as a cluster of servers. What is a cluster? A cluster is a group of servers that perform the same function in parallel. In this sense, we can have multiple web servers or multiple data servers that distribute the workloads of a firm amongst themselves. The fact that in order to increase capacity, I do not have to make my single server larger, but I can rather add more servers to my architecture is what makes a cluster of servers particularly scalable. There are several ways in which I can organize this cluster of servers. For example, we can have what we call as a primary server and secondary or slave servers that will only come online if the first server happened to come down. Sometimes this is referred to as a cold backup in the sense that we usually only have one server running, and only if this primary server comes down, its slave or its secondary server comes online. We may also have what we call a hot redundancy, case in which we have two or more servers concurrently running performing the same functions. In this case, we would have servers functioning as mirrors of each other. So in this regard, we say that having a cluster will increase the availability of the servers, not by making a single server more powerful or more available, but rather by having redundancy across different servers. If we were to draw a cluster of servers on a network diagram, the icon we could use is one as shown on the bottom right part of the screens. Moving to our final type of servers, we can speak of virtual servers. And what is a virtual server? In layman's terms, we could describe it as a server within a server. And what do we mean by this? Let's take a big physical host. This could be a high availability server or it could be a mainframe. And let's install a software tool called a hypervisor on this server. And within this hypervisor or on top of this hypervisor, we're going to be running multiple virtual machines. So how does this work? If you look at this in terms of the IT stack, trying to understand what do we need to run a software application. We first need hardware, CPUs, memory and storage devices that are part of any server. And on this hardware, we're going to install the operating system. So for example, we could install Linux on top of our hardware. Now, within the Linux, we will install what we call a hypervisor. And what is a hypervisor? It is a software tool that allows us to segment or partition the entire server in multiple portions and then create virtual servers or smaller sized servers within this hypervisor. And we can have multiple virtual servers running within a hypervisor. The most common hypervisor in corporations is that offered by VMware, and in particular VMware ESXi. This software is traditionally used by corporations and it has been out since the late 90s, I believe, to run virtual servers within servers. A competition of VMware, and that in this case is also available as open source, is Send Server. Send Server has become particularly popular in the cloud context. Because Citrix offers a more refined version of Send Server that is the one used by cloud infrastructure providers such as Amazon or Rackspace. Recall, cloud infrastructure service providers offer virtual servers to their customers on their public clouds. Another example of hypervisors you might be familiar with is the software used in laptops and in particular Mac computers to run a Windows within the Mac computers. VMware has an offering designed for laptops called VMware Fusion, but other solutions competition of VMware's Fusion are Oracle's VirtualBox and Parallels. These are softwares that we're going to install on our laptops that run one operating system, and we install this software to run another alternate operating system within our laptops. Let's move to another element of the local AIR network. We're done with servers, let's start talking about clients. What are clients? Clients are the devices that access the servers. Thus, they are the hardware used to input and output information by final users, and they're going to be the means by which we access the servers or other clients. Examples of clients are going to be the standard PCs and laptops we see every day. Now, back in the day in which we only had mainframes, we could talk about dumb terminals. A dumb terminal was nothing more than a screen on a keyboard that we would use in the late '80s to connect to a mainframe. Basically, users would have no computing power at their desks, but rather all the computing power was hosted at the mainframes, and all we had on our desk was a keyboard and a screen to interact with the mainframe computer. Not too far from that, we have what we nowadays call network computers. Network computers don't have too much processing power, but they have just enough power to run a virtualized desktop that is running off another server somewhere in the world. So, all what we have is a very slim device on our desk that allows us to see the screen of a virtual desktop running elsewhere in the world. Another example of clients are what we call transactional terminals. These are clients designed to perform a very specific function. If you think about it, the bank's ATMs, which we use to withdraw cash, have a single function, which is served to connect the physical customer with the bank's information systems. Another type of thin client that we might use for a single function is a point-of-sale. These point-of-sale devices do not need to be powerful computers, since all the functions they perform are those related to the checkout process at a retail store. And finally, we have all our mobile devices, starting from smartphones, moving on to tablets, watches, wearables, basically anything that can move. So, note that the term "client" refers to anything that is connected to a network, and that in turn can be used by the user to connect to services offered by servers. Let's move on to our final set of devices, namely the network devices that interconnect all the servers and clients. And let's talk about switches, access point routers, and modems. Let's start with access points. Access points are these devices you might have seen on walls or ceilings, and that have antennas. These antennas are the means by which devices such as our smartphones or laptops that also have antennas connect to a local air network. We will basically have our laptops interacting with an access point, and the access point is in turn connected to the local air network of a firm, and thus through this wireless device we are able to access all the services in the local air network. Now, some of you might have seen these devices in your apartments or homes. These devices are kind of tricky, because they do have antennas, so they function as access points, but at the same time they interconnect all the devices in the apartment, which is the function of a switch, and they also route the traffic of our apartment or house to the outside network, in this case our internet service provider's network, or the internet. So these devices, sometimes manufactured by Linksys, D-Link, or other manufacturers, are a sort of 3-in-1 combo that has an access point, a switch, on a router. So let me talk specifically about switches and routers independently to clarify what these do. A switch is nothing more than a junction point where cables come in and cables come out. The smallest version of these might be a very small 8-port switch. And what's gonna happen here is that data coming through any of the individual ports of the switch will go out by any of the other ports in the switch. So it's basically, again, a junction point for everyone to communicate with everyone else connected to a switch. In a firm, however, we will not find such small switches. Rather, we might find switches with 24 or 48 ports that are very slim, and they are slim because we're going to organize them in the racks we spoke about when we talked about data centers. And this is how a series of switches might look at a rack on a particular data center. So once again, access points allow us to connect to the local air network via wireless radio signals, and switches interconnect all the devices within the network. Now, once we want to go outside of the network, we need to route the traffic from a local air network onto or towards other networks. And for this, we use a router. A router is going to be the device that interconnects different local air networks. And the ones I'm showing you right now might be routers used by a firm. However, if you're a larger corporation, in particular an internet service provider, the router might look like the size of an entire cabinet or an entire rack. This is the router that might be used by a cable company or an internet service provider. And that's it for the network devices. But how do we connect network devices with each other and with service and clients? Let's now talk about the communication media. What we typically know as a network cable is also known as an Ethernet cable or UTP cable. UTP stands for unshielded twisted pair. And we refer to twisted pairs for the little twirlings of copper cables we see in the cable. The jacks of these cables look very similar to the traditional phone jacks. And all the laptops might have network ports to which we can connect one of these cables. Although laptops generally nowadays only connect through wireless means. Another type of copper cable used to transmit data are copper coaxial cables. Through these cables we can cover much larger distance than with the UTP cables mentioned before. You might have seen them as the cables used for televisions. And basically you can transmit data through the same cables through which we can transmit analog TV signals. Also if your internet service provider works with cable modems, this is the kind of cable that will be reaching you. It's a coaxial cable. More and more we need faster speeds and more reliable cables. And this is what has driven the usage of fiber optic cables. Rather than transmitting electric signals to the copper, we're now transmitting light through the fiber optic cables. The two main benefits of using fiber optic cables are speed and distance. We can cover much greater distances with fiber optic cables than we can with copper. And also we can transmit data. We can transmit light at much faster speeds. Moving on to wireless. Perhaps the moment I said wireless, the first thing you thought of was Wi-Fi hotspot. A Wi-Fi hotspot is nothing more than an access point. That allows us to connect to a network, in particular local layer network. Sometimes we connect to a hotspot to access the internet. However, we must understand that before accessing the internet, we're really connecting to the local layer network of whatever place we are. And from there, a router takes us from the local layer network onto the internet. That being said, there is much more wireless communication media than just Wi-Fi. For example, take cellular networks, 3G, 4G, and whatever there is to come. Cellular network antennas look like the ones in the picture. And they are a form of wireless communication. Another type of wireless communication is through microwaves. And antennas look like the ones in the photo up on the right side. Communication through microwave antennas is characterized by needing line of sight. Meaning that you must have a clear, straight line from one antenna to another to have direct communication. We'll talk more about these when we mention metropolitan area networks in just a second. And as a final example of wireless communication, I want to mention satellite communications. These, of course, go all the way into outer space, but they also qualify as wireless communication. So far we have mentioned a wide array of devices and servers. And perhaps it is best at this point to make a pause in introducing new concepts and talk to you about an example. The example I want to describe to you is very close to many of those watching this video. And it is a quick tour inside the small data center we have in our Costa Rica campus here at Incai. Let's watch the video. So we're entering into our data center. You can see it's locked. You can see all the ethernet cables. These are patched ports that are going all throughout campus. These devices down here are the switches through which all the traffic of our network passes. Down here we have fiber optic cables that route our backbone network. We can see we are also connected to the internet through fiber optic cables. We have further networking equipment on our second rack here. Several servers, routers, firewalls, switches. And we finally have the UPS, the uninterrupted power supplies, the batteries, which are turned on in case we lose main power here into the data center. We have several servers which are rack mounted servers. And we also have a more old school server working on its own, which is a standalone server. We can see temperature is maintained very cool, around 20 degrees Celsius. And something very important about this location is that it has its independent electricity supply. A final topic I want to address on this video is the type of networks we can talk about based on their geographic extension. And what I mean by geographic extension is going to become clear as we go through the examples. I'm going to be talking about local area networks, backbone networks, metropolitan area networks, and wide area networks. So let's start with the local area network, which is the one we've been talking about since earlier in the video. A local area network is going to occupy the space of a room or a building. It's going to have all the elements we've talked about, namely switches, routers, servers, clients, printers, all the ones we've been talking about. They interconnect everything within a room or within a building. Typical speeds of these networks are going to range between the 100 megabits per second and the 1 gigabit per second. They're meant for nearby communication between devices. But in a firm, we can have multiple buildings. With the example on the right, I'm inviting you to imagine a firm that has several buildings. In one building, they have their manufacturing facility. In another building, they have all the people working on maintenance, all their stock and inventories, historian warehouses. And they have another building just for the administrative staff. And we need a mechanism to interconnect all these buildings with each other. The network we're going to be using to interconnect the buildings is what we would call a backbone network. The scale of a backbone network is going to be less than a few kilometers. The elements that compose or that build up the backbone network are going to be local air networks and the devices we need to interconnect the local air networks within each building. So we will be having here high-speed switches and routers, and we will also have high-speed circuits. Circuit is the formal name we give to links within a network. So when you talk about high-speed circuits, we will quite likely be referring to fiber optic cables used to interconnect the different buildings at high data rates. The typical speeds of a backbone network will range between the 1 to 40 gigabits per second. Moving to an even broader extension, we can talk about a firm having multiple branches, multiple locations within, say, a city or within a metropolitan area. Think of a bank that has multiple branches scattered throughout a city. The scale of the metropolitan area network is then going to be more than a few kilometers. What we will have is different buildings, different facilities actually, interconnected by circuits, that in this case, the circuits are going to be quite likely leased to public providers such as a cable company. Why don't we use our own fiber optic cables? Well, because it is very expensive to throw fiber optic cables on our own privately-owned cable from one location to another, which is 10, 20, 30 kilometers away. Rather than laying out our own fiber optic cable, it is less expensive to actually lease these circuits to lease the fiber optic cables of cable companies or telecom providers that have already laid out their own fiber optic networks through the cities. An alternative, if we don't want to lease a fiber optic link, or if we want to have a redundant link, is to have point-to-point connections through microwave antennas. Remember, these will only be feasible if we have line of sight, which means a straight line between one branch and another without any mountains or big forests in between. These kind of links, regardless if we use microwaves or if we use leased circuits from public providers, are generally going to be expensive. And sometimes, rather than interconnecting ourselves directly through a private metropolitan area network, we might use an internet-based tunnel or channel, perhaps through a VPN tunnel, that relies on the internet as the way to communicate between one branch and another. Typical speeds for these kind of networks are going to range between the 64 kilobits and the 10 gigabits. You might wonder why 64 kilobits, which is a very, very low data rate. But this is the lowest data rate you can use to have a voice over IP conversation. So sometimes you want to have a network just for voice communications, and 64 kilobits is going to be more than necessary than what you need for this. However, these are less and less common, and we might talk about links that have at least a few hundred megabytes of capacity. Extending the concepts of a metropolitan area network, but to much larger scales, and perhaps even international scales, we talk about wide area networks. Wide area networks are going to be the private networks used to interconnect multiple operations across the globe for a single firm. You can think of a South American logistics provider that has multiple branches on different countries. That firm will likely have a wide area network through which all the different firms in the different countries interconnect with each other. So once again, the concepts are actually very similar to those of our metropolitan area network, particularly in the sense that we will quite likely be leasing our own links unless we have our own private satellite-based network. And the speeds of this network are going to, again, range between 64 kilobits and 10 gigabits per second. Two further ways of referring to networks you might have heard of are intranets and extranets. These are nothing more than another way of classifying networks, but now based on who can access them. The names are fairly intuitive, but let me elaborate briefly on them. An intranet is a network accessible only by internal members of an organization. So these are standard collaborators. And for example, everyone connecting in a local area network to the ERP system, to the central information system, is going to be connecting to it through what we would call an intranet. So in general, everyone connected to the local area network, or even to our wide area network, is going to be part of our intranet. However, physical presence on the premises of a firm is not necessarily needed to be able to connect to a firm's intranet. It is quite common that collaborators who work remotely, be it working from home or on a business trip and at a hotel, they might use a VPN, a virtual private network, to connect to the firm's intranet. We will not, however, discuss what a VPN is in this video, since that is more a security-related topic. We finally have extranets, which are networks designed to be accessible by people or entities external to an organization. A very widely known example of a big extranet is how Walmart offers its suppliers access to its stock, such that the suppliers themselves know when it is necessary for them to start shipping more goods into Walmart's warehouses. They know this by connecting to Walmart's extranet. A perhaps more common example is simply how we, as end users, as customers, go into a firm's e-commerce website, such as Amazon.com or any other firm that offers us an online portal. If you think about it, the online portal is part of that firm's enterprise systems, where we have information on prices and stocks and we can place purchase orders. So that e-commerce website could be categorized or classified as part of the firm's extranet. We may also find connections not necessarily into systems, but just into networks, and the public Wi-Fi offered by a retail store for its customers could also be called an extranet. And with that, we conclude this video. I am certain that many of you have many more questions about the different elements of IT infrastructure discussed in this video. However, recall that the scope of this video was to educate business executives and give them an overview of the different elements in an IT infrastructure, rather than teaching executives how to administer. That is the job of the system administrator. This is German Retana from Incai, and I thank you very much for your attention.