Yet Another Linux Blog

I have spent the last couple of hours using Ubuntu Breezy (5.10) on a Celeron 400MHz machine with 256MB RAM, 13GB HD, DVD-ROM and jeez is it slow! Opening a new tab in Firefox takes a couple of seconds, having a few tabs open makes everything crawl, switching apps takes a few seconds, etc. If I try installing a package and switch to Firefox, it’s even slower.

I have to admit, I’m used to fast hardware and don’t have much patience for computers. I always want my computer to be faster than me; if i hit Ctrl-T, lxer.com, Enter, Ctrl-T, slashdot.org, Enter, Ctrl-T, news.google.com, Enter, Ctrl-Tab, I certainly shouldn’t have to wait in between the new tabs. Even as I’m typing this post, the letters appear after a certain delay that just drives me nuts!

I tried installing the Opera Browser because it’s supposed to be a bit faster than Firefox on the same hardware, but I haven’t found that to be true at all. They are qualitatively equally “snappy” on this hardware. Opening a new tab is still slower than it should be. Perhaps an older version like 7.0 or 6.0 would be faster.

On the positive side, I would consider this machine to be usable by the average person. This machine can do everything a faster machine can do with Ubuntu, it’ll just take a few minutes more. Sound and video works, so does flash (OK, so I had to add the multiverse repositories and know to install mozilla-flashplayer and gstreamer-plugins). I think it’s better to suffer through slightly slow Firefox or Opera than to suffer with the limitations of a browser like Dillo or links.

There is clearly a need for Ubuntu Lite. The original target for that sub-distribution was to configure an Ubuntu distribution to be usable on a machine with less than 64MB RAM. I have until recently been a proponent of the idea that RAM matters much more than CPU (that is, a 200MHz Pentium with 256MB of RAM would be far “faster” for desktop use than a 400MHz Pentium 2 with 128MB RAM) I based this opinion on my experience of using an Athlon XP 2400+ machine with 64MB RAM (by using the mem=64m kernel option (but also a fast ATA100 disk)), and it was much “snappier” than I expected. My experiences with this machine (400MHz, 256MB RAM) really make me doubt my previous opinion.

In either case, it’s much harder to choose the right packages to result in a “usable” desktop when you only have 32MB RAM. In fact, the latest Xorg needs at least that much memory. That’s why I think Ubuntu Lite should really target less than 256MB machines. There seems to be some widespread agreement that GNOME is a bit “heavy” and certainly, with a low-resource machine it doesn’t make sense to devote so much memory and processing power to the desktop environment itself. The resources should be left to the apps.

Anandtech has a nice article testing the latest dual-core Xeons vs. the latest dual-core Opterons. They promise to include power measurements in their benchmarks from now on.

The summary: Xeons perform as well or up to 15% better than the Opterons in this benchmark but the Opterons use half the power at full load! If you pay for electricity, you probably want an AMD-based system.

Or is he?

I started out writing this post with the full intent to say some unkind words about a fellow named George Ou who blogs over at ZDNet. I wanted to talk about how he’s a Microsoft shill and how he consistently offers his biased opinions as “facts”. But perhaps my own vision is colored by my views on Free Software. In either case, look for yourself: Could Desktop Linux really be so slow? or Performance analysis of OpenOffice and MS Office. Take a minute to read through the comments that question his testing methodology and through his replies where he accuses the person of being biased and offers further unsubstatiated opinion on the state of Open Source software.

George has a large number of blog posts detailing his “testing” of OO.org with invariably the same conclusion: it’s slow, it’s “bloated”, and it sucks. He compares it to MS Office and looks at things like memory footprint and load times as the benchmarks. I would personally think that the most important performance factor is the features that you need to use, not the amount of memory it takes up or the number of seconds for it to start up. You can buy a faster computer for less than what it would cost you to buy MS Office.

George does redeem himself somewhat with this post: Why can’t Microsoft just patch everything? where he laments about the number of unpatched vulnerabilities in MS Windows and IE.

The Five Rules of Cool is an editorial about Apple’s five strengths. I think these ideas are really tenets that vendors can follow to be successful. For example, one could build an “Ubuntu box” and partner with AMD.

After all, Apple has more marketshare than Linux, so they’re doing something right.

In a recent Slashdot story about Thunderbird, we had a small discussion about the superiority of webmail. I’m currently a Gmail user, mostly because I use multiple computers throughout the day. Even if I did use just one computer, I still don’t see any advantage to using a local mail program.

Webmail pros:

• accessible from anywhere, anytime
• unlimited space (for my average needs, anyway)
• protection against data loss

Webmail cons:

• accessible from anywhere, anytime, if you have a network connection
• unlimited space (for my average needs, anyway), unless you need 3GB of e-mail
• protection against data loss, if you trust Google to do backups
• doesn’t work well on PDAs
• can’t check more than one account at once

What’s clear is that webmail is not ideal for everyone. If you use your PDA all the time, the Gmail interface doesn’t scale well to your tiny browser window, and you’re better off with a mail client. If you don’t trust Google to keep all your mail (I think they do backups more often than I do), you should keep a local copy. If you’re not always connected to the network (I am), then you need a local copy of your mail. If you really use multiple e-mail accounts( I just forward them all to Gmail), then you need a mail client.

Of course, even if you do have special needs as listed above, you can still use Gmail, and just use their POP3 interface. The guys at Google really have done their best to enable everyone to use their awesome service.

I realize this post sounds almost like a Google advertisement, but I really didn’t mean it that way. I just like Gmail because it offers me all the features I need. And I really like the search functionality. And the automatic spam filter. And the fact that when there is a tracking number in the e-mail body, Google recognizes it and puts a link to the tracking information in the sidebar. And the fact that it nicely hides the quoted text in top-posted mail. And the fact that they add more and more features over time.

One of the critical things needed for Linux world domination is a good replacement for Microsoft Exchange. Exchange is the leader in what’s called “groupware”, or software that allows groups to communicate and organize better. “Groupware” typically includes integration between e-mail, calendaring, task lists and contact lists.

One competitor is Citadel, described in this blog entry.

An even better page is here: Migration Alternatives For Microsoft Exchange 5.5. Microsoft’s support for Exchange 5.5 is expiring on January 1st, 2006, which is only a couple of weeks away. The linked page has a good list of alternative software that can do the same thing as Exchange.

Paul Murphy at ZDNet argues that Apple should release the OS X GUI for Linux.

The trend seems to be the following:

• Apple succeeds at something and makes a large profit.
• Some people really like Apple’s stuff but don’t want to pay for it.
• Those people talk about how much money Apple would make if they made their products cheaper/more available

A few posts ago on this blog, there was a comment suggesting that Apple license OS X to other hardware manufacturers. This time, Paul Murphy wants Apple to make their code available on a Linux platform. Guess what people? OS X is Apple’s advantage over everyone else. It is the leverage they use to sell their hardware. They are a profit-maximizing firm, and they’re pretty good at it. They have 7 billion dollars in the bank, and they’re not going to cannibalize their hardware sales by letting you have OS X for free/cheap.

So if you want to use OS X, shell out the cash, and quit whining.

This post is a business idea. Ok, maybe not so much a business idea as a website idea. It is spurred by my previous post about the need for qualified, paid Linux support. There are a great deal of people who want help with their Linux systems, or would want help with their Linux system if they knew they could get it. There are a number of people who are highly qualified to provide such support. What’s missing is the connection.
Sure, IBM has a bunch of smart Linux guys. Sure, Sun has a bunch of smart Linux guys. Sure, Red Hat has a bunch of smart Linux guys. But they’re really targeting the high-end enterprise, the people who are willing to pay the same license fees for Red Hat “advanced server” as for MS Windows 2003 Server.
What we need is a community of qualified Linux consultants, all committed to the same goal, all willing to go one-on-one with any of the “average Joes” that need help with their computers. And we need an excellent way for the “average Joe” to get help from such a consultant.
The problems with this idea are the same as with any computer consulting business that is targeted at the home consumer. The home consumers are not very friendly customers, oftern too annoying or irrational for a qualified person to deal with. They also aren’t willing to pay very much. The best effort I’ve been so far in this regard is the Best Buy Geek Squad. The problem with the Geek Squad is that Best Buy simply cannot afford to pay them enough to hire well-qualified people, mostly because people aren’t willing to pay much for the service. When a new Linksys wireless network costs you $75 for the hardware, you don’t want to pay another $75 for some guy to install it for you. When you ‘apt-get’ the latest software for free, you don’t want to pay a guy $100 to configure it properly.
An interesting point to consider is that Free Software eliminates one of the cost factors from the computing process. You no longer pay for the software itself, but you still need to pay for maintenance and tech support. If you could also cut out the middleman from the Geek Squad service and pay Linux experts directly, you eliminate yet more cost.

This is a draft from 2005-12:

Expensive backups software:
Veritas Backup
Retrospect Backup
Arcserve Backup

I have listed the above software in the order of their popularity. Many people who use Veritas like it (except for the price). Many people who use Arcserve do not like it. Retrospect is so-so.

There are a number of Free software solutions. My favorite, and thus the one I will highlight here is called BackupPC…

update in 2008! I’ve been using Bacula at work as well as BakBone NetBackup. One is free, the other is expensive, but both do an excellent job. The user mailing list for Bacula is top-notch!