..bits & bytes teleported

I had the opportunity to review the article -Exploring Scalable Data Processing with Apache Hadoop-  by Tom Wheeler, which is now published.

My article on Apache Hadoop: Hands-on Hadoop for cluster computing is now live at http://www.linux.com/feature/150395

              +                 

• Set up 3 Virtual Machines
• Set up internal networking between the hosts
• Set up SSH-less login among the hosts
  
• Extract Hadoop into the same location in each Virtual Machine
• Follow this document at http://hadoop.apache.org/core/docs/current/cluster_setup.html to set up a Hadoop cluster in a box!
  

http://www.google.com/search2001.html

Scenario: You need a internal network among your guest machines. Internal network would mean a sandboxed network among your guests.

Solution:

In the user manual, installing the guest additions is well documented. However, where do you get the guest additions ISO?

On Linux it lives in /usr/share/virtualbox (provided you have installed it to the standard locations)

PS: Guest additions provide a lot of value addition to your guests. So, if you are looking to do anything serious, go ahead and install them. I have installed the guest additions for Windows XP and Solaris. Works like charm.

It is well documented that Google Chrome has a multi-process architecture and forks a new Chrome process for a every new TAB or a plugin. 

I have got 4 tabs and 1 plugin running:

 

 Here is the process tree: (using Process Monitor)

 

 The visual explain the multi-process hierarchy. During startup there is a browser process which forks a new renderer process everytime a new tab is opened and/or a new plugin process everytime a plugin is opened.

The blog post Multi-process Architecture is an excellent read.

 I am working on an article on Google Chrome and have been exploring the various usability/UI features and the internal architecture. 

In terms of the User Interface, IMO Google Chrome scores with the following features : (in order of my likeness)

1. Application Shortcuts: Using this feature you can place a Desktop/Start Menu shortcut to a web application of your choice. Whenever you launch the web app using the shortcut, you will not see any of the browser features- Back/Forward/Reload buttons and misc other browser features. All you see is the web application. In a sense the website starts behaving like a true web application on the desktop and does not give the impression of running in a browser. I like this feature since I often have applications like GMail running all-day and all I need is the web application and not the browser with all its features. More on Application shortcuts here
2. No Crab like mentality : If 1 out of the 10 tabs open crashes, its the ONLY one. The other 9 are usable and safe. No crab like mentality
3. Task Manager : Google Chrome has a multi-process architecture at its core and has an inbuilt task manager that lists the various processes currently running under the browser. Think the core browser to be your OS and the open tabs, plugins as processes. More here
4. Dynamic Tabs: With tabs as the primary focus, Google Chrome has truly dynamic tabs enabling a seamless tab based browsing experience. More here

All the features are listed here at http://www.google.com/chrome/intl/en/features.html

A must-watch video is the Google Chrome announcement: http://www.youtube.com/watch?v=LRmrMiOWdfc&feature=user

Check this link at http://www.linuxjournal.com/content/google-chrome-linux to try Google Chrome on Linux.

This is a proof-of-concept magic using WINE by CodeWeavers

The VirtualBox Images project at http://sourceforge.net/projects/virtualboximage gives away pre-configured VirtualBox images of various Linux/Free BSD operating systems - downloadable from http://virtualbox.wordpress.com/images along with a script to automatically register the new image in VirtualBox.

Gotcha:

The images can only be used on the Hosts on which they were created- this is a property of VirtualBox. The only way to find this is the error message that you will get when you attempt to register the image. It would have been nice if there was a mention somewhere.

For example, The DSL image can only be used on Mac OSX hosts as it was created on one!

Follow a related discussion here at http://vbox.innotek.de/pipermail/vbox-users/2008-September/003906.html

I am going to try ScribeFire as my blogging client...

Primary Link: http://www.virtualbox.org

Accessing the host operating system from a guest is as simple at setting up a network adapter attached to NAT and you will be able to access the Internet, ssh to the host and other things.

So how does this work?

The guest gets a IP address via DHCP which is made possible by the internal DHCP server of VirtualBox. The host also gets a IP address which is only accessible by the guest. We can think of it to be a internal IP address. Hence, we can “talk” to the host OS.

However, we cannot communicate in the same way with the guest OS from the host OS. This is expected as this is the very concept of NAT

Solution: Host Interface Networking

From the VirtualBox user manual:

“With Host Interface Networking, VirtualBox creates a new networking interface in software on the host computer. This new software interface will then exist in parallel to your regular interfaces (e.g., on a Linux host vbox0 will exist alongside eth0). When a guest is using such a new software interface, it looks to the host system as though the guest were physically connected to the interface using a network cable: the host can send data to the guest through that interface and receive data from it. This means that you can set up routing or bridging between the guest and the rest of your network.”

Set up a Network bridge to enable Host Interface Networking

The VirtualBox user manual has detailed instructions for setting up host interface networking on Windows, Linux and Solaris.

Here I will reproduce the steps for Ubuntu Linux 8.04.1

• Install the bridge-utils package: sudo apt-get install bridge-utils


• Edit your /etc/network/interfaces to add the following lines:
  



auto br0 <br />iface br0 inet static <br />address 10.10.3.4 <br /> bridge_ports eth0 <br />



• 
  Restart the networkin services: sudo /etc/init.d/networking restart


• Create a software host interface: sudo VBoxAddIF vbox0 amit br0 (where 'amit' should be the user who will be running VirtualBox)
  


• Modify the network settings of your virtual machine to add the interface 'vbox0' : VBoxManage modifyvm JeOS -hostifdev1 vbox0 (where 'JeOS' is your VM name)


• Boot up the VM and assign a static IP adress to the interface
  
  

Now you can access the Guest from the Host and vice-versa.

For eg. You can have a webserver running on your Guest OS and you can acces it from your Host.

 

All virtualization products use their own formats for the virtual appliances due to which virtual appliances created using a particular product can only be used properly with that specific solution. This is not a happy situation.

Open Standards to the rescue

A new development which promises to overcome this short coming is a new standard for packaging virtual machines- Open Virtualization Format (OVF) (http://www.vmware.com/appliances/learn/ovf.html)

It was perceived by the Distributed Management Task Force (DMTF)

OVF, among other things will allow interoperability among the various virtualization products availalble.

Compared to VMDK or VDI which encloses only a single virtual disk in the virtual machine, the OVF format provides a complete specification of the virtual machine. This includes the full list of required virtual disks plus the required virtual hardware configuration, including CPU, memory, networking, and storage.

The OVF is a standards-based, portable format that allows the user to deploy this virtual machine in any hypervisor that supports OVF.

OVF makes heavy use of XML and the technical specifications are available at http://www.vmware.com/pdf/ovf_spec_draft.pdf

More information on OVF is available at http://www.vmware.com/appliances/learn/ovf.html

ovftool

The ovftool is a Java based experimental tool to convert VMs to and from OVF, and converting standard sparse/flat VMDK files to and from the compressed stream- optimized VMDK format used in OVFs. (VMDK is the file format used by VMWare for virtual disks)

It is available for download here at http://www.vmware.com/download/eula/ovf_eula.html

...I say a LOT.

For those of you who use GMail, there is a 'Sponsored Links' section on the right side of your screen which displays relevant-to-your inbox kind of advertisements. (http://mail.google.com/mail/help/about_privacy.html#targeted_ads  )

Normal stuff.

Today I received a mail from a friend who wrote Bengali in English- that is Bengali words written in English. A snippet:

And here are the sponsored links I saw in the right-column:

So, did GMail actually "learn" that it was Bengali written in English and hence produced links that were 'Bengali' in nature?

Amazing.

Communications of the ACM (August 2008) has an article on 'Games with a purpose: GWAP'. 

By involving human beings in such fun and interesting games, game designers leverage human knowledge to make computers more intelligent.

Few places to visit:

• http://www.gwap.com/
  
• http://images.google.com/imagelabeler/