So, whilst going through my daily fix of Slashdot , I chance upon this . I read it as Solaris/x86 has poor driver support. I guess Linus couldn’t very well say that Solaris virtual memory subsystem sucks . I’ve heard similar dismissive comments from vendors during a Linux conference in HK and it’s sad that whilst Sun is opening the source, people aren’t opening their minds.
Whilst its true that Solaris x86 supports fewer hardware than other open source operating systems like Linux/FreeBSD/NetBSD/OpenBSD, what it does support it supports quite well. I for one would be very grateful if Solaris/x86 were to support the 3ware raid controller cards. Solaris 10 + ZFS and with Sun’s robust NFS server implementation would mean that Solaris owns the build-your-own NAS market. 3ware is supposedly looking into this
Sun has supported a lot of large open source applications including Gnome, OpenOffice and Mozilla. I am most familiar with the Mozilla developer community so I will write a bit about what I see missing from Sun wrt Mozilla community.
For those new to Mozilla development, Mozilla developers hang out on irc.mozilla.org in various channels. Won’t list them here.
There is always some developer or the other on the channel generally talking about the codebase, answering users question (even if they are in the developer channel) etc. I see engineers from Redhat, IBM and other organizations who have a vested interest in Mozilla on that channel. However, I haven’t come across Sun engineers on that channel unless they are having some issues with a commit they just did and the tree is burning.
Sun has a fairly large team in Beijing (all hail our mainland masters ) and it should be possible that like IBM/Redhat engineers they also mingle with others and not work in isolated silos. I’ve mentioned this to Henry Jia sometime back. Is it a cultural issue ? I don’t know.
Similarly, Sun could loan a pre-configured Solaris 10 box with Sun Studio to Mozilla Foundation and provide a walkthrough with Dtrace and libumem. If either or both of these technologies helps Mozilla performance, then its great for the Mozilla community and reflects well upon Sun.
Sun is looking for a community to sprout around OpenSolaris. Their behaviour with existing open source communities will shape the perception of potential OpenSolaris community members.
Solaris is not just the kernel. It’s a combination of the kernel and the desktop environment. The kernel group has done phenomenal work in getting the message out about Solaris 10 technologies such as Dtrace, Zones, SMF and FireEngine.
The OpenOffice/Mozilla/Gnome groups in Sun need to do the same. You see Dan Williams of Redhat responding to users at osnews.com in the discussion about OpenOffice 2.0 Preview . I’d expect someone from Sun to be also there clarifying misconceptions and communicating the larger picture.
I was chatting with Alvaro Lopez Ortega who works in Sun Ireland a few days back and mentioned to him about event ports and how I felt that would be great to have in his high performance web-server Cherokee . He hadn’t heard about it . I told him to drink the Solaris kool-aid and get cracking on better support for Solaris with Cherokee. But before that Alvaro has to do the autoconf magic to get Cherokee to compile with Fedora Core 3.
Maybe there is a lot of low hanging fruit to pick in terms of internal evangalizing which Sun needs to do.
According to the South China Morning Post SCMP (subscription required) , two more mobile phone operators launched third-generation (3G)services on Tuesday – but only for customers who had already ordered handsets. Anyone going into a CSL or SmarTone shop hoping to sign up on the first day of service was asked to put down a deposit on handsets – for delivery after Christmas.
As to lack of content, my previous post mentions the need for Apple Hong Kong to get more aggressive in evangalizing Quicktime
Hong Kong has a mobile subscriber penetration of 117 % . I think this has to be a world record
Now, Apple has always maintained that its Quicktime technologies is very well suited for 3G networks. However, Quicktime is not listed as an option in the webcasts offered by the Hong Kong SAR Government
These webcasts are only in Real Media and Windows Media player format. I was speaking with someone recently who setup a streaming media system for another government dept where they used Windows Media and asked him why not Quicktime and he said that he hadn’t heard about it.
In my opinion, Apple should evangalize Quicktime streaming more strongly in HongKong, the demographics of the population is such that they are receptive of cool technology, there is a fairly sizable music/entertainment industry (John Woo,Jackie Chan, Chow Yun Fat, Maggie Chueng just to name a few) here to provide content and the government itself promotes streaming media (albeit not Quicktime at the moment).
With the potential increase in 3G subscribers, I feel that Apple has the unique position of gaining substantial mindshare amongst the mobile phone/PDA user segment which may increase Apple’s market share in desktop. I guess it would also help if Apple Hong Kong provided discount to students similar to the Apple Education store in the US but that’s something for another day.
Masood Mortazavi blogs about the doubling of mobile phones since 2000.
Joi Ito also blogged about how China eats mobile phones for breakfast
Here’s some info from a Kyocera investor presentation
In order to revitalize the image of PHS in the market, we launced
handsets equipped with an Opera browser that enables access to the
Internet in the same way as a PC. These handsets also have the benefit
of a flat-rate for continuous Internet connection via AirH
Since its launch in May, stores have been selling out of this type of
handsets due to its popularity.
I am a bit confused about what people mean when they say “WAP is really big these days”. Is it the old/clunky WAP/WML that was available in 2000/2001 where most sites had a special version written in WML ? Or does WAP these days refer to a web-browser (typically Opera) on a cell-phone where the user is interacting via HTML over HTTP ?
Darin ‘Necko God’ Fischer mentioned to me via email that Bug 220941 would be a good bug to fix for better performance of Mozilla over wireless links. He also mentioned that it’s one of the more difficult bugs to fix hence could be one of the reason the bug hasn’t caught much attention.
Speaking of wireless, in the current Linux 2.6.x kernel (as of 2.6.9), there are now 3 congestion control algorithms available.
- TCP Vegas
- TCP Westwood
- BIC TCP (default as of 2.6.9)
My understanding of TCP Westwood is that it is a sender-side modification of the TCP congestion window algorithm that improves upon the performance of TCP Reno in wired as well as wireless networks.
From the abstract of the ACM-Mobicom 2001 paper I gathered this.
Unlike TCP Reno, it doesn’t blindly halve the congestion window after receiving three duplicate ACK’s. Instead it uses the current effective bandwidth (TCPW continuously measures the TCP source rate) to set a slow start threshold. In wireless links, one could have sporadic losses due to radio problems and TCP Reno would interpret this as congesion which would lead to unneccasary window reduction.
So with a lot of cellphone manufacturers looking at Linux on the cell-phone, will they tune the stack for TCP Westwood. Not sure if other OS’s like Solaris 10 support different congestion control algorithms. Can one set congestion algorithms on a per-route basis ?
Bryan ‘Performance is thy middle Name’ Ryner has a patch lined up in bugzilla which gives a 5% increase in page-load on Linux (not sure if the win is on all platforms).
Looks like this will be available if Seamonkey/Firefox is compiled with the gcc packaged in Fedora Core 3. Since Solaris 10 build 73 will also come with gcc 3.4.x, I am not sure if it has the same patches which bryner mentions as being required. I’ll try to ping some Sun engineers about this. It could be that Sun’s native compiler (Sun Studio 10) may be doing the right thing
Found this page which describes a bit more about the effect of the symbol visibility patch in GCC.
Sun Engineer Kyle Yuan shows his mozilla-fu by blogging a bit about how one would use dtrace to grovel through the beast that is Mozilla. Looking forward to seeing some bugzilla reports of performance improvements and memory leaks being plugged due to Dtrace.
Thanks to Sun’s crypto team for donating an optimised AMD-64 assembly implementation for the bignum library. Bug 272327 has the details.