Friday, October 26, 2007
Spliced Networks upgrades to Leopard
Here at Spliced Networks we use Fedora Core Linux workstations based on AMD Athlon64 X2 hardware, and multiple LCDs. However we use Apple Macbook and Macbook Pros for our mobile needs. We have a number of Mac Pro servers for development on the MacOS X platform as well. Today, we've moved the MacOS X systems to Leopard, without any problems.
Friday, October 12, 2007
Linux.com feels the heat
It seems that the traditional media is starting to feel the heat from o3 magazine's high quality, high tech content. Many of you were aware that we signed on Mayank Sharma as Editor in Chief. Mayank is a great young guy from India who works as a freelance open source journalist and editor. Mayank was being eased into the role of EiC, he has done a great job editing o3 articles and he just started writing for o3 with issue 9.
Following a week long visit from Linux.com Robin "Roblimo" Miller down in India, Mayank was forced by Linux.com to quit o3 magazine. We wish Mayank the best of luck with his future ventures. We already have two new editors signed up, but we are reworking some of the articles that Mayank had access to for issue 10.
We have decided to not run issue 11 on Ohio LinuxFest, as we felt the coverage over on our o3 linuxfest blog was sufficient. Instead, issue 11 will look at JeOS - "Just Enough OS". Both issues are on their way..
Following a week long visit from Linux.com Robin "Roblimo" Miller down in India, Mayank was forced by Linux.com to quit o3 magazine. We wish Mayank the best of luck with his future ventures. We already have two new editors signed up, but we are reworking some of the articles that Mayank had access to for issue 10.
We have decided to not run issue 11 on Ohio LinuxFest, as we felt the coverage over on our o3 linuxfest blog was sufficient. Instead, issue 11 will look at JeOS - "Just Enough OS". Both issues are on their way..
Thursday, October 11, 2007
Spliced Networks adds 200MBit/sec in Europe
Spliced Networks has added multiple servers behind a 200MBit/sec connection at a new data center location in Germany. The new location will operate as the primary EU-NOC for Spliced Networks. The new location is expected to be rolled into production within the next week or so. Spliced Networks is expanding its resources in San Jose over the next week.
rPath down again for several hours
About a month ago rBuilder suffered from a rolling repository outage. The infrastructure problems continued this morning when large parts of rPath's infrastructure went down. This may or may not be related to the hour of maintenance conducted between 9pm and 10pm EST on rBuilder Online yesterday. But before 4.30am this morning (EST) large parts of rPath's web services were down according to Antonio Meireles, due to a bad proxy. Services were restored around 7.21am.
A quick look at rPath's infrastructure looks like it is just some colo sitting on some Cogent bandwidth. Nothing wrong with Cogent, we use Cogent in some places, but we're multi-homed and multi-site. A quick DNS lookup on www.rpath.com shows its on 38.100.0.24, the first IP in this block used for rPath appears to be 38.100.0.19 (colo-admin.rpath.com), with 38.100.0.28 appearing to be the last used (at least with configured reverse dns). Hardly enterprise grade, no redundancy, perhaps they should read o3 magazine! :)
Live logs from #conary:
[05:56:22] doniphon > large parts of rPath web down. bad proxy, etc.
[05:59:00] iwilson_ > I saw that
[05:59:11] iwilson_ > ironically enough, they die when I'm searching for something.
[05:59:16] tpfennig > oh downloads are slow today...
...
[06:06:14] doniphon > tpfennig: rBO is flacky atm. anyway fill a bug. *that* should not happen atm.
...
[07:09:35] doniphon > msw Up2 mkj jtate SM2k gxti *@rPath. wiki is down
[07:16:56] SM2k > doniphon: we're working on it
...
[07:21:19] msw > back///
A quick look at rPath's infrastructure looks like it is just some colo sitting on some Cogent bandwidth. Nothing wrong with Cogent, we use Cogent in some places, but we're multi-homed and multi-site. A quick DNS lookup on www.rpath.com shows its on 38.100.0.24, the first IP in this block used for rPath appears to be 38.100.0.19 (colo-admin.rpath.com), with 38.100.0.28 appearing to be the last used (at least with configured reverse dns). Hardly enterprise grade, no redundancy, perhaps they should read o3 magazine! :)
Live logs from #conary:
[05:56:22] doniphon > large parts of rPath web down. bad proxy, etc.
[05:59:00] iwilson_ > I saw that
[05:59:11] iwilson_ > ironically enough, they die when I'm searching for something.
[05:59:16] tpfennig > oh downloads are slow today...
...
[06:06:14] doniphon > tpfennig: rBO is flacky atm. anyway fill a bug. *that* should not happen atm.
...
[07:09:35] doniphon > msw Up2 mkj jtate SM2k gxti *@rPath. wiki is down
[07:16:56] SM2k > doniphon: we're working on it
...
[07:21:19] msw > back///
Wednesday, October 10, 2007
Utility computing is going to cost you..
A quick browse of Web Hosting Talk's dedicated hosting offers forum, and you'll see that you can lease a nicely configured Intel Core Duo server, 1GB ram and 160GB of disk space on a dedicated unmetered 10MBit/sec link for around $130. Thats full duplex too, so about 3255 GB of transfer in each direction a month. Assuming we're not taxing the server too much, lets assume 20% CPU utilization. Plugging this same data into a utility computing service such as Amazon's S3/EC2, we end up with a bill over $590!! Prior to June 2007, it would have been almost $1000. So I could double up at the same data center ($260), and put two servers at another data center in Europe ($200), and still be UNDER the cost of an Amazon EC2 service by over $100. Not to mention that I've also got 4x the capacity. So this model really only makes sense if you can't administer your own server and like to throw away money!!
Subscribe to:
Posts (Atom)