VIT’s very own Cloud Computing Centre #3

It’s now time for the third article in the series of articles on cloud computing.

This time I will cover Eucalyptus, a Linux based , open-source software infrastructure for the implementation of cloud computing on computer clusters which provides an interface that is compatible with the Amazon EC2 service.

A Eucalyptus cloud setup consists of five types of components. The cloud controller (CLC) and "Walrus" are top-level components, with one of each in a cloud installation. The cloud controller is a Java program that offers EC2-compatible SOAP and "Query" interfaces, as well as a Web interface to the outside world. In addition to handling incoming requests, the cloud controller performs high-level resource scheduling and system accounting. Walrus, also written in Java, implements bucket-based storage, which is available outside and inside a cloud through S3-compatible SOAP and REST interfaces.

Top-level components can aggregate resources from multiple clusters (i.e., collections of nodes sharing a LAN segment, possibly residing behind a firewall). Each cluster needs a cluster controller (CC) for cluster-level scheduling and network control and a "storage controller" (SC) for EBS-style block-based storage. The two cluster-level components would typically be deployed on the head-node of a cluster. Finally, every node with a hypervisor will need a node controller (NC) for controlling the hypervisor. CC and NC are written in C and deployed as Web services inside Apache; the SC is written in Java. Communication among these components takes place over SOAP with WS-security.

Many instructions in this article refer to a single-cluster installation, in which all components except NC are co-located on one machine, which we refer to as front-end. All other machines, running only NCs, will be referred to as nodes. In more advanced configurations, such as those with multiple CCs or with Walrus deployed separately, the front-end will refer to just the machine running the CLC.

Eucalyptus can be installed from source or using a set of packages (RPM and DEB). The former method is more general and should work on practically any Linux system, the latter is easier but will only work on the distributions that Eucalyptus support. As of 1.6 they are:

CentOS 5.4,
Debian squeeze,
OpenSUSE 11, and
Ubuntu 9.04 "Jaunty" and 9.10 "Karmic".

To install eucalyptus on CentOS refer this link : http://open.eucalyptus.com/wiki/EucalyptusInstallationCentos_v1.6

For configuration details refer the following links in order:

http://open.eucalyptus.com/wiki/first-time-setup-16
http://open.eucalyptus.com/wiki/hypervisor-configuration
http://open.eucalyptus.com/wiki/EucalyptusNetworking_v1.6
Once Eucalyptus is installed, the management tools called Euca2ools should be installed. For this refer to http://open.eucalyptus.com/wiki/Euca2oolsCentosInstall_v1.1

To troubleshoot, look at the eucalyptus forums.  You can also ask your questions by commenting on the article.

Next time, we will look into cloud management using Euca2ools.

VIT’s very own Cloud Computing Centre #2

This is the second instalment in the series of articles on Cloud Computing. As promised in the previous article, this article will cover two major topics

 

 

 

• Deployment Models
• Importance of Interoperability and open standards in Cloud Computing.

Deployment Models:

1. Public Clouds: Public clouds are based on the standard cloud computing model, where a service provider provides resources (Applications, platforms, Infrastructure) to the general public over the Internet. Public cloud services may be free or priced on pay-per-usage model.

Benefits:

1. Easy access to resources through the Internet
2. Inexpensive due to the payment model.
3. Scalability(The most important benefit of cloud computing i.e. elasticity)

Examples of Public Clouds: Amazon EC2 service –> paid service ->provider of Iaas, Google App engine –> paid service –>PaaS, Google Docs –>Free (I'm not sure if they have an enterprise subscription) ->SaaS

2. Private Clouds: Private clouds are similar to Public clouds, except that they are build on private networks. The idea behind private cloud is to deliver the benefits of cloud computing at the same time reducing the security risks that arise with public clouds. Private clouds are best suited for large corporations that are not comfortable with having their data on a public cloud. It is estimated that private clouds will be adopted more in the coming years until all the security concerns are addressed. Eucalyptus is an open source software used to build private clouds. The cloud infrastructure that is being build at VIT by the Linux User Group is powered by Eucalyptus.

3. Hybrid Clouds: Hybrid cloud is an environment in which some resources are kept in-house (may be on a private cloud) and some resources are placed on the public cloud. This setup is also ideal for corporations.

Interoperability:

Interoperability w.r.t Cloud Computing refers to the ability of customers to use the same artefacts (Management tools, OS images, APIs) with a variety of cloud computing vendors. Let me give you two examples:

a) GNU/Linux images running on Amazon EC2 should be able to run on Slicehost without changes.

b) The Office documents created on Google Docs should be editable on other Vendor services.

The idea behind open standards and interoperability is to avoid vendor lock-in so that customers can migrate between vendors depending on their service. It is an important issues as proprietary lock-in can hinder the adoption of cloud computing. If cloud computing is to move ahead, vendors need to put aside their differences and agree on common principles related to security and the interoperability of cloud platforms. A lot of groups and forums have been formed to address these issues of which the important ones are CCIF and DMTF.

In the next article I will cover a step by step procedure to build your own Private Cloud. Till we meet again, Happy reading.

Collaborative MindMap – Anand Narayan

As we promised earlier, we have a special article by Anand Narayan himself sharing his experience as the developer of Collaborative MindMap: and interface you can use to collaborate online with like minded people. The idea is ingenious and we’re proud to have an article from Anand himself on our website.

Anand is pursuing his B. Tech., CSE from VITU and he is currently in the third year.

As an introduction, we have his video about Collaborative MindMap here:

And without further delay, we have the man himself. Here’s what he feels:

5 semesters over, 3 to go. I wanted to build a nice app that I would fetch me some money before I get out of college. So I brainstormed with a lot of my connections, did some market research, finally landed up at one. Collaboration was (is) the in thing. Even Google (wave) was into it. And one of my mentors suggested I do a collaborative mind mapping tool. Sounded pretty interesting, something I could sell. So we collaborated and charted out the features and I started with the development.
Step one of every project; you got to choose the technology you are going to use to implement the idea. I asked a lot of people, social networks, friends, geeks. Didn't get many suggestions. Ended up with a couple of libraries, stuck to one. Learnt it for a week. Then had no clue how to use it to develop what I wanted, felt like I was stuck with too ambitious a project. This is where it really helps having a mentor, he broke it down, and I finally got going.

It was full code now, after like 3-4 days of coding, I just finished connecting a couple of rectangles and stuff. I realized that I had chosen a very low level technology, which meant that I had to write a lot of code and in-return I get a lot of power. For instances I had to code even a simple textbox and trust me it’s not that easy to code a textbox, loads of events have to be taken care of at such a low level . And so I continued and had a lot of fun. I was able to appreciate computer graphics; so much of code goes into them, browsers, desktop interfaces and stuff ... Lovely.

After sometime with development, I reached the crucial collaborative part. I had a naive idea about how to go about this, again googling, Wikipedia, only made things worse. The more I read about them, the more I got confused. Fortunately I landed on some code and I got the concept in like minutes. It’s so weird theory is always confusing, but it is actually meant to make things simpler.

So now my working prototype was almost ready, I just had to test for a few bugs, and I was ready to go. I made a screen cast and showed it to a few people and got some feedbacks. Eventually I attended Barcamp Chennai, a few of the people whom I had shown the idea to, were present there and they really egged me on to demo my prototype there. Thanks to them I did speak at Barcamp and I demoed the product, got some nice supportive positive encouragement from the people there. They gave good suggestions and I really felt like I should do more to this product, because so many people like it, Felt great. I was wondering how Steve Jobs would feel when he shows off a product like iPad to the whole world and have them jaw drop .

So this where I am now , I am looking at developing the UI , adding more features , making an alpha version , then beta , then release . I'm also looking for people who can work along with me. If you are interested mail me at anand[at]sanandnarayan.com

We hope you enjoyed this inspired you to do something similar. We’d love your feedback, you can always follow us on Twitter: www.twitter.com/VITreporter or just email us: contactus[at]vitreporter.com

We’ll be back with more!

Special thanks to Aditya Vikram Thoomati for letting us use his image.

Top Programming Myths

Each one of us often wonders - how do we become good programmers? And that question becomes much more predominant if you’re going to get an engineering degree within a period of four years.

First, let us shatter some misconceptions. The first one is what most students believe – the more programming languages I learn, the better I will become. Wrong. In fact, you couldn’t be more wrong. Programming is like poetry. And when you try a different programming language, it becomes more or less the same situation where you switch to completely different writing language and try to come up with meaningful poems. In fact, a new programming language will only result in the overhead of adapting to the syntactical changes.

The second of most prevalent misconceptions – I should learn programming from a book. Let us ask you a question from the perspective of poems again – which poet do you know of, learnt poetry from an English grammar book? Again, programming is something best learnt through experience. Yes, a reference book is definitely going to help, but it is not going to be a source of new ideas in terms of algorithms. A programming language book is only going to explain to you the grammar of the language, the syntax. It will not inculcate in you the skill to come up with an algorithmic solution to a problem.

The third – those who took programming in their high school can only be good at it. Again, false. Programming is like mathematics and poetry combined. You can learn and master programming at any stage in life, given you move in the right direction with it.

Okay, enough myth-busting. Let’s get down to ‘how to learn programming’. The key my friends, is experience combined with a sharp analytical skill. First, one needs to learn how to effectively translate and break every situation down into a programming problem. Almost everything encountered in engineering can be put in terms of programming. Once you’re able to do that, half the work is done.

The next stage is the planning and analysis. Hold your horses before you reach out for the keyboard and begin coding. First, get yourself some old fashioned piece of paper and a pencil. Now, analyse the problem statement mathematically. Take some cases, try them out, see how the math flows. Next, move on to the more extreme cases and see if the flow of things is still the same. If not, repeat the previous step, until you begin to get the drift of things.

Now comes the part where your computer and the compiler are your best friends. You’ve done the planning, you’ve done the analysis, it’s now time to sit down, and code. In real programmer vocabulary, this part is not as exciting as the planning and analysis part. Ask any good programmer and you’ll find out.

So what do we conclude from this? Firstly, you need to know just one programming language. Yes, classical computer science doesn’t need you to know more than one. A programming language is just a way of translating mathematics into a computer’s language.

Secondly, try programming more for mathematical situations rather than things like sockets and other such utilitarian pieces of code. Unless you’re really interested in run of the mill Information Technology, you don’t need to know much about such code.

Thirdly, spend more time with a pencil and a paper than an IDE. Imagine more, and think in terms of mathematics.

Fourthly, get an algorithmic bent of mind. Instead of issuing a reference book from the library, buy yourself a copy of a hardcore algorithmic bible like “Introduction to Algorithms” by CLRS or “The Art of Computer Programming” by Donald E. Knuth.

Lastly, patience. Which of course, all of you who have read till this line, have.

Cheers and happy coding.