Om Malik?s Broadband Blog » Long Term Impact of Voice over IM

The Internet connection here didn’t improve much even at after midnight, so it wasn’t the peak hour traffic that was a problem… I called LodgeNet and the guy had no clue and told me he would work on it. I was tired anyway so I went to sleep. Woke up this morning , same thing. Called them again, and I connected my PowerBook to the ethernet directly just so he could ping it. Turn out the Airport Express was the problem. So I thanked him and opened up Airport Admin Util, and I noticed the Airport Express itself has a 10.10 IP address, the hotel DHCP server also assigns 10.10 address, so that didn’t look right. So I pretty much enabled the Distribute IP addresses feature in the Airport Express, which basically turns on its DHCP server, and I had it distribute 198.168 address… rebooted the Airport Express and wala, fast Internet connection again! 🙂

The hotel must’ve changed something yesterday, as the previous Airport Express configuration worked before, and I’ve been here for more than 10 days. hmm…

The connection at my hotel is insanely slow at peak hours, like right now, 20:00 PST… Crazy packet loss… 🙁
[PowerBert:~] ayn% ping www.google.com
PING www.l.google.com (64.233.179.104): 56 data bytes
64 bytes from 64.233.179.104: icmp_seq=0 ttl=240 time=136.066 ms
64 bytes from 64.233.179.104: icmp_seq=2 ttl=240 time=1929.158 ms
64 bytes from 64.233.179.104: icmp_seq=3 ttl=240 time=929.055 ms
64 bytes from 64.233.179.104: icmp_seq=4 ttl=240 time=140.096 ms
64 bytes from 64.233.179.104: icmp_seq=5 ttl=240 time=1066.102 ms
64 bytes from 64.233.179.104: icmp_seq=9 ttl=240 time=9098.939 ms
64 bytes from 64.233.179.104: icmp_seq=10 ttl=240 time=8099.114 ms

I think I’m gonna watch my NetFlix DVDs instead, I can’t stand slow Internet connection.
Continue reading →

Venture Chronicles by Jeff Nolan: Other Shoe on Google Talk

Of course Jeff Nolan was right, but the thing is, there are already Jabber/XMPP transports that work with AIM/ICQ, MSN, and Yahoo. I’ve used Jabber for years and they work just fine. You do need an account on all the other services to talk to your friends. But Google might add features in their client on top of the regular XML-based Jabber, so you can convince your friends to use their chat client, and then after a while they might very well switch to using Google if the voice (or video, or whatever) feature only works there.

The problem with Jabber is that it was never easy to use, and it was never very stable. To register my Jabber account on my friend Dan‘s server I had to telnet to a certain port at the server and type in XML code (now most clients support registrations). jabberd, the Jabber server, was sorta a hack, and seems like nobody cared enough to maintain it. The MSN and Yahoo transports were also not maintained. With Microsoft and Yahoo trying to block other clients from using their services every other day, they didn’t work half of the times. This was true until very recently, I’m not sure what happened, maybe Dan installed newer versions of the server and transports. Jabber has worked flawlessly for the last several months.

Also, if you domain name doesn’t resolve to your Jabber server, you’re pretty much screwed. And I wanted to have ayn@mydomainname.com instead of ayn@jabber.mydomainname.com. I had the vision that Jabber will eventually replace all the IMs and it would work more like emails, with distributed servers that handle IMs. The Jabber specs supports _jabber._tcp SRV DNS records, so in theory if I include a line in my DNS entry to indicate to where Jabber TCP traffic should go (sort of like the MX entries), it should work. But I don’t think it ever did, last time I tried was 3 years ago though, so this might not be true now.

Google, once again, took an existing technology, vastly improved it (I’m sure they rewrote most of jabberd), added features to it (voice), and rolled it out as a brand new service. Probably 90% of the Google Talk users had not heard of Jabber before. This is pretty cool.

Jealous lover program writer nicked

LOL!

How to Predict Google?s Next Product

Google Talk

Nice, DIE AIM, Yahoo IM, and MSN… Actually, I’m still gonna stick with Skype unless Google talk is a lot better… (with features like encryption, SkypeOut, SkypeIn, and voicemail)…

Looks like Google Talk is nothing but Jabber, I’ve used Jabber for about 5 years now, with my friend Dan’s server… Oh well…

UPDATE: The Windoze client works for me, but I can’t get to it with Adium via Jabber, weird…

Interesting, Chinese low-cost ASIC design engineer body shop, the day is near… 🙂

VeriSilicon – Profile:
HQ: SHANGHAI, China Founded: 2001 Management: CEO is Wayne Dai, who was the Co-Chairman and CTO of Celestry Technologies, which was acquired by Cadence Design Systems in 2002. Prior to that, he was the founder, Chairman, and CEO of Ultima Interconnect Technology. He has a Ph.D. degree in Electrical Engineering, from UC Berkeley and was a professor at UC Santa Cruz. Investors: In August 2005, Verisilicon closed Series B with US$13.5M. New investors included Intel Capital, HSBC Private Equity (Asia), CID Group, Legend Capital, KTB/UCI, International Finance Corporation (IFC). Existing investors were WI Harper Group, IDG Technology Venture Investment, iGlobe Partners, and Harbinger Ventures. Business Model: Verisilicon is an ASIC design foundry; aka – a low-cost chip engineer body shop. Services range from a design and verification platform, IP, front-end and back-end design services, software design service and turn-key services for China based foundries. VeriSilicon had 140 engineers prior to its latest round of funding. Competitors: Faraday Technology, Paradigm Works, ViASIC. Dirt: Even as Intel has added more and more functionality to its chips, the demand for specialized chips continues unabated. VeriSilicon has proven that it can do good work in China – and for much less than shops…

So as most of you noticed I had problems with the server and it was down for a day, I’m not exactly sure what was wrong. But /dev/hda3 got errors and automatically remounted as read-only per /etc/fstab, I tried `fsck -y /dev/hda3`, it finished, rebooted, server came up but sshd (or any other services) didn’t come back up.

Submitted a trouble ticket to ServerBeach and they put in a new hard drive and re-generated the server, with the old hard drive in there as well (though they forgot to mount it). I suspect it was a hard drive failure, but I can never really know… Took them a day to get it up though. Trevor and I spent about 3.5 hours getting everything back up and running.

I got another server from ServerMatrix, if it works out well I will be migrating the stuff over in the following weeks. Here’s the server stats:

• Server: Super Server 2.8 GHz
• Primary HDD: 80GB Hard Drive
• Secondary HDD: 80 GB Hard Drive
• Drive Controller: IDE
• RAM: 1024 MB RAM
• Number of ips: 5 IP Addresses
• Bandwidth: 1200 GB Bandwidth
• Uplink Port Speed: 100 Mbps Uplink
• Web Analytics: Urchin 100 profiles
• Database: None
• Backup Service: NAS 20GB
• Operating System: Debian (stable)
• Server Management Plan: Silver
• Control Panel: None
• Firewall: None
• VPS Software: None
• Salesperson:
• Special Requests or Comments: Sarge version

So my tenants just moved in, they wanted wireless so I got them another Airport Express… I setup mine as a range extender, pretty pointless, but cool: