Pagi ini tiba di kantor, dan tiba-tiba ditegur teman, “are you coming to this OpenLDAP meeting ?”. Hm, perasaan meeting hari ini soal desktop linux deh, tapi oh well, why not.

Ternyata benar bukan soal OpenLDAP, tapi pengenalan soal Open Source dan Open Desktop oleh netproject. netproject adalah perusahaan konsultansi yang menjadi terkenal ketika Microsoft kebakaran jenggot karena Newham City Council hampir pindah dari solusi Microsoft ke solusi open source berkat rekomendasi dari netproject. netproject juga membuat laporan untuk European Union soal cara migrasi ke solusi open source.

Eddie Bleasdale, direktur netproject, memimpin langsung rapat tersebut. Berikut adalah beberapa hal yang menarik dari rapat ini:

1. Proposal paten software di EU (European Union) ternyata adalah karena tekanan keras dari pemerintah Amerika dasar teroris.
2. Kasus migrasi open source di Munich – ternyata keputusan migrasi tersebut berdasarkan hal yang politis, not based on good business/technical case. Menurut Eddie, saat ini mulai terdengar beberapa masalah dengan implementasi pada proyek ini. Duh, mudah-mudahan mereka tetap berhasil deh…
3. pak Eddie pernah sekali memberikan presentasi mengenai open source ke para manager marketing Oracle se-Eropa; dan mereka semua sepakat bahwa MySQL (bukan DB2, bukan juga MS-SQL) adalah ancaman nomor #1 bagi Oracle

Cukup menarik….

Juga ada beberapa hal yang menarik yang lainnya, tapi sayang belum bisa saya beberkan sekarang. Mudah-mudahan dalam waktu yang tidak terlalu lama lagi deh.

… BIG TIME

Why ?

1. Reliability: This thing takes IBM literally years to stabilise. I personally enjoying it myself, with my manager calling me “why the website is down ?”, and to find out that Websphere has fell down again for no clear reason. And we’re only using it for running our CMS for God’s sake (think Mambo, Drupal, PostNuke, etc) – it’s not something terribly sophisticated ! In fact, that CMS is actually developed in Tomcat ! So it’s not even using the full feature of J2EE offered by Websphere. Still, Websphere is having problems running this
2. Forced upgrade: It all started pleasantly – the IBM consultants ran the excellent presentation on Websphere, and the price was reasonable too. Fast forwards several months – we’re having reliability issues with it. We inquired IBM about this, and their reply was some sort of “well, we no longer support that old version (what !!), so if you want to get our support, you better upgrade it to the latest one”. Then we found out the nasty surprise of the upgrade cost…
3. Cost : Latest versions of Websphere are so expensive, once it was priced 8x (eight times) higher than the alternative solution based on Tomcat. Basically, picture yourself a big, bad, Sun server; with multiple UltraSPARC processors, and gigabytes of RAM. OK, basically, the Websphere license cost was enough to buy 2 of those. Do I still need to tell you which one got chosen by the project manager ?
4. Scalability: An analyst defended Websphere saying that it’s the right choice when you need scalability in a system. Sorry, but, scalability my a$$. Basically, despite given the best server in the system, Websphere consistently managed to still become the bottleneck of that system. I ran a set of load-tests a few months ago against a system, which include a Compaq DL380 and a big Sun server running Websphere. That Compaq DL380 runs 3 virtual servers (using VMware) , yet it’s load consistently hovered around 1%-2%. The Sun server (that runs Websphere) however, kept on falling over flat on its face, with 100% system load.
5. A bloated pig: I think by now you’ll understand already that Websphere is basically a massive bloated software, with low performance level compared against other similar software, despite attempts by (very expensive!) IBM consultants to tweak its performance.

It’s a massive bloat, it’s slow, it’s unreliable, and it’s priced to burn a seriously big hole in your wallet.

Stay as far away as possible from it.

UPDATE:

# There’s a reason why Microsoft picked Websphere in their effort to promote .Net against Java
# Finally IBM will revamp Websphere’s reliability and performance. Or is it possible that I’m rejoicing too soon ?

OpenBSD has been released for quite a while now. [ Here ] is a nice summary of it from Newsforge.com

My personal impressions of it are:

1. It’s number #1 OS if you’re concerned about security
2. There are so much work going into this OS – hardened against buffer-overflow attack (even built into its compiler), arguably the most common type of attack. Also impressive use of cryptography, and how the system is set up securely in general – no initial root login via ssh, chroot-ed daemons, etc.
3. OpenBSD team gives the best impression of a professional voluntary group. They set high standards, they set goals, and they meet them.
4. An OpenBSD server, however, can be really difficult to setup and maintain. This is more to our lack of familiarity to various security process, instead of the fault with OpenBSD itself. Still, it tend to deter newcomers.
5. Another problem is lack of newbie-friendly documentation. I tried to rectify that with [ this ], however note that I had to stop writing it when I changed my server before I can really sure that everything in there works.
6. Package management is a serious omission in OpenBSD. But I heard that such thing (some kind of portupgrade) is already availabe in the bleeding-edge version of it.
7. OpenBSD doesn’t yet comprehensively address the new rising threat – web-application security. Including mod_security in Apache by default could help to address this significantly

My own biggest issue with it is lack of effective package management solution (which is important when you want to spend as little time as possible administering your servers), so once this is addressed (in v3.7 I assume), then I think I’ll give this another go.