First Dive into OpenSolaris

These few days I have been playing and experimenting with OpenSolaris 2008.05 and Ubuntu 8.04.

At the surface, OpenSolaris looks very similar to Ubuntu or any distro of Linux that runs gnome:

• It also provides most of the open source tools that shipped with Linux
• The look and feel are 'almost' the same
• It has the authentic Unix directory file structures and design philosophy which Linux borrows from

The first difference I noticed is the incomplete functionality of the OpenSolaris UI (or 'frontend' in software engineering jargon) . Some of the tasks can only be done at the command line.

For example, the UI only provides button for the user to shutdown the system, but there is no 'restart' button. If you want to restart the system, you need to open a text terminal, and type 'reboot'.

Another example will be the network setting. When OpenSolaris boots up, it has the Network Auto-magic daemon (nwand) running. If your network settings are unusual or you want to manually do some settings, you have to disable the nwan and enable the 'default' settings manually. When you click on the 'networking' icon, it will just ask you the network cannot be set while nwan is running and you have to disable nwan first. It then points you to read nwan(1m) and let you figure out the rest.

Guess where you can set all these? Command line of course.

svcadm disable svc:/network/physical:nwam
svcadm enable svc:/network/physical:default

Since the shell will accept shortest unique string, lazy smart user will do:

svcadm disable nwam
svcadm enable physical:default

For those of us who are used to text terminals and commands, OpenSolaris is something that we know and love. In fact, often command line is desirable since it is often more responsive, and tasks are easier to automate.

The 2008.05 release represents a good progress for OpenSolaris and I feel the underlying backend is reasonably stable. The remaining work for Sun and the OpenSolaris community will be:

• Port more packages to the OpenSolaris platform
• Enhance the functionalities of the frontend without sacrificing the versatility provided by the underlying infrastructure (i.e. tool chain and command sets)

One of the main attractions of OpenSolaris is the renown ZettaByte File System (zfs) which is really really sexy. According to Sun, zfs has self-healing and redundancy built right in the file system and this could easily make hardware backup solutions like RAID or tapes obsolete. If the claim is true, we are seeing yet another example where software slowly replaces hardware functionalities.

I have a zfs set up (one 80GB hdd) and now managed to export it as nfs. My plan will be get Samba to run on it and thus creating a central repository for both Windoze and Linux boxes.

Stay tuned.

Safari on Windoze

Very recently, Apple started its next foray into the lucrative PC market after migrating to the x86 platform last year. This time it ported its web browser, Safari on to the Windoze platform for everybody to try on. (Link is here)

Since I always like Apple products, I downloaded Safari for a test drive. The result is very disappointing though.

To clarify, I am reviewing this product from an ordinary user's perspective, using it just for normal day-to-day chores: online banking, web surfing, blogging, etc. No attempt was made to break it (more on that later). Even though this is still in the Public Beta phase, I am expecting all the normal things to work.

Upon clicking on it, the launch speed is indeed impressive. However the catch is it fetches the initial web site much slower than its peer (relative to firefox, but still faster than IE, obviously). I guess the Apple engineers just launch the skeleton _first_, such that it is fast as claimed, but doing all the initializations right after that.

I also tried to import some of my firefox bookmarks, but the import bookmark feature just didn't work, basically it just did nothing after I specified the file. On top of all these, online banking also didn't work and Safari just killed itself after I keyed in my username and password.
After all these trials, I decide to just leave it around until a more usable version comes out. For now, I still have my good faithful firefox. Using IE? You are kidding. :P

Not surprisingly, Safari already has a security vulnerability discovered (see here). So much for the 'world's best browser'.

The Negated Database

The Economist has an interesting article on a new cryptography idea: instead of securing the data with encryption that may be compromised, why don't we just store the negation of the data? i.e. The database contains everything except the intended data (if the data space is small, we may need to put some extra paddings to increase the entropy). Say a database of 8-digit phone numbers, that database will contain all 8-digit numbers, except those valid ones. So to check whether a phone number is valid, we need to make sure that number is not in the database.

The advantage of this could be the leakage of confidential data may be contained if the said database is breached. However, I can think of the following shortcomings:

1. Any change of data formats (e.g. adding one more field) will warrant a respawning of the whole negative database. This will not be trivial if the data space is big enough.
   
2. Searching and sorting will be a challenge as special algorithms must be in place to ensure the database is responsive enough.

I feel the underlying theory has something to do with entropy: This negative database has a much higher entropy than the normal ones, and therefore it will take more effort to deal with it (this applies for legitimate and illegitimate users!).

Update:
Author's homepage.
He has a paper with more information about this scheme, however it seems my two concerns are still not addressed...