Server/Network Architecture & hello, AMIH!

I'm new here and the employees love it. I am "The Rookie".



It's day one, and I've attend the daily morning meeting (to discuss yesterday's systems and applications and today's plan) as an observer. The meeting has seven participants, and contains a mixture of employees from AMIH and (the latter handles the raw coding for the software that AMIH needs to run its IT operations). In the meeting are specialists in the following areas: heterogenous computing, network architecture, server architecture, enterprise resource planning (specifically ), and the software development life cycle.




I am brimming with questions. I need clarification for the multitudinous swath of acronyms being tossed across the conference table. Some of these are clarified above. I've come to realize IT people are acronymophiles.



The time I spend with AMIH is to 1. learn about the activities of an IT company (what better place to do so than India, to where IT stuff is outsourced) and 2. write a leave management system software application. I start off with objective 1.



Following the morning meeting, the network architect (Satish "The Undertaker") and the server architect (Ashutosh "Early Bird" - who's always early to work and late to leave)and I ("The Rookie" - by the way I give these nicknames to my colleagues and we use them with each other!) go to another conference room to discuss their specialities - with the idea of giving me a layman's understanding of server/network architecture. Let's talk about networks architecture and firewalls first. Then let's tackle server architecture.



FIGURE 1: A POINT TO POINT LINK BETWEEN HQ (WHERE I AM WORKING) AND A REGIONAL OFFICE AT TUTTHUKUDI (TUT)



The basics.



provides the virtual pipelines sustaining the point to point link between HQ and the regional office. In this way, Reliance is to network architecture as Wipro is to applications programming (software development) for AMIH. The (local area network - remember playing Counter Strike 1.5 over LAN? Wasn't that fun! Same thing here, mang.) between HQ and TUT is extended by , which is a family of methods used to control network traffic to create virtual private networks (VPNs).



I'm throwing a lot of information at you here. But bear with me.



Network Topology



FIGURE 2: NETWORK TOPOLOGY USED BY AMIH IS PRETTY STANDARD ACROSS IT COMPANIES



HQ, TUT are data centres, connected via point to point link provided by the Reliance MPLS VPN Internet service provider (). Regional offices 1 and 2 can connect, through the ISP, to HQ and TUT.



The Firewall



The firewall determines what sites you can access. These sites are recognized by ports. A port is basically a communications endpoint. There is a unique IP address / port number for every computer in the world. Wow this is boring. Lemme throw in a picture for ya.



FIGURE 3: WEEE A PICTURE (FIREWALLS 'N STUFF)



The nebulous looking cloud is the great unknown the outside.



End users are typically on the "outside" of a network especially in company like AMIH, which deals with heterogenous computing primarily. Note heterogenous computing is the synthesis of many systems and applications to bring a wide variety of data together. These end users are seldom on the inside of a network where the (more down-to-earth and concrete) coders dwell. End users are a little more abstract than the coders (like the ones AMIH employs from Wipro) in that they do not directly implement/code the applications they use. And not all those on the outside can access what's in



"There are 0 to 55355 ports available" mentions The Undertaker. He's a big guy, like the wrestler. The name fits and it's origins I'll explain in a bit. The firewalls around the data centers (HQ, TUT) decide which ports you want to access from the inside and which ones you don't want to access. No unauthorized user (supposing there's a connection) can access the inside if the firewall is present. Bam. Safety.



Firewalls create VPNs (sorry for the acronymophing. I'm an IT guy now ) so you can access the inside from the outside.



There are three types of accessibility we are worried about



1.



"Extended LAN" so to speak. For example, our HQ to TUT connexion. Client access for a single user.



This is the one we're most worried about. *



2. Client access



I'm not going to explain this. Muahahaha. Go it.



3.



"Layer 2 Tunneling Protocol". A tunneling protocol that simply requires a connexion to pass through a tunnel to ensure privacy. This type of accessibility isn't so secure it provides too easy access.



* So how does this site to site thing work?



There are two ways HQ and access TUT and vice versa. Site to site occurs over the internet. But if there's no internet connection, HQ and TUT can establish a connexion using MPLS VPN over the secure network provided by the ISP dude, Reliance. Nice.



Switch to server architecture



FIGURE 3: STANDARD SERVER ARCHITECTURE



Basic idea for server architecture.



Here's a bit of analysis based on Figure 3. User 1 can access an application (Human Resources Management System or HRMS, for example, is an application used by AMIH) through the central switch, which provides a LAN. This central switch is also a data centre. Switches in general are used to distribute LAN. This central switch is connected to servers 1 and 2. Say the HRMS app is stored on server 1. Then user 1 accesses server 1 through the central switch to run the HRMS app.



Supposing there's some user 2 who wants to access the HRMS app, but they aren't connected to the central switch. Then they must come through another switch, like switch 2 (switch n depending on how many n switches you want) in your system. The path is then: switch 2, central switch, server 1.



THE UNDERTAKER



"Do you watch WWE wrestling?" asks Satish all of a sudden.



"Uhhh when I was a kid a little " I say.



"Because a server is like The Undertaker" says Satish. "While it takes 1 guy to bring down Rey Misterio, a small wrestler, it takes like 10 guys to bring down The Undertaker" who, like Satish, is a huge guy. That's why servers host applications, which take up a ton of memory.



TIME FOR SOME COOL SERVER FACTS.



Within a server, you can define policies to control who can access which apps.



There are two types of servers. (Graphical User Interface, or "gooey")-based and nonGUI-based servers. GUI-based servers are Windows servers, typically. These include Windows 2003, 2008, and 2012 servers. nonGUI-based servers are unix servers hpux, solaris, aix, and linux are examples. I was surprised to learn that even though Apple looks GUI, it's based upon unix so Apple computer servers fall into the nonGUI-based category. Weird



That's all for server/network architecture, folks.
Full Post