Good morning,
Today is monday the 22nd of January 2018. The time is 04:21:06 and it's week number 04.

(2009-05-10) The ERICADE Network-NG turns one!

One year has passed since we switched from Windows to Linux and got real hardware to run it on and it's clear from the evidence it was a good plan.

Challenges ahead
As we stand today, the system is working better than ever before. The load on the system i very light and we have no resource problems what so ever.

We have a spare system in case of disaster that is not yet fully operational but can be switched to on a short notice.

The up time is high despite a few problems with the power in the first two months. We actively monitor the SMTP, WWW and DNS-service and will know about any failures within minutes.

A new major release of Kerio Mail server is around the corner and we plan a system upgrade around the time it gets released. This will probably be done in early June during the night. More information on this will be published on this site. Onwards!

A bit of history:
In 2005 I installed a mail server on a private PC and moved my own mail to it. That's how Secure ERICADE Net got started. It was meant to be a testing ground for my own interests.

But things changed rather quickly when the mail server hosting began falling apart. As a stopgap measure I moved all my domains to my own system. And what was meant to be a temporary solution became permanent.

In 2008 the system was working well but I had a few problems with the hardware that was not built to be on line 24 hours. The whole system was built with desktop PCs. The main server was a VMWare server hosting the mail, file and directory servers as guests. This worked fairly well, but it took 20-25 minutes to reboot all systems and it had serious problems with memory consumtion.

In may 2008 I decided to rebuild the whole solution from scratch. I replaced the main server PC with a real HP Proliant server that was actually built to work 24/7. Windows 2003 server was replaced by Fedora Linux and Exchange was replaced by Kerio Mail server.

Stability and security was tremendously improved and the occasional reboot could be done within 3 minutes with no manual procedures.

In July I let Pingdom start monitoring the systems and record its up time. The goal set was that we should have no more than about 16 hours of down time per year. Or in other terms: 99.8% up time. We also set a rule that there should be no more than 1 hour and 20 minutes of downtime per month.

In the fall it was obvious that everything worked fine, except one thing: our ISP. Bredbandsbolaget had numerous breaks in their service, but most of the time it was during the wee hours so we believe nobody was ever affected. The uptime goal was 99.8% and we managed to reach 99.78% which is close enough to feel we actually did ok. Most of the downtime was due to the ISP.

One nasty incident was when our firewall failed in October 2008. Thanks to preparedness we had a backup firewall that kept the system up while we replaced the faulty one.

January and February came with a nasty surprise. We had three major power failures in the area which caused disruptions and racked up a totalt of 10 hours downtime during the first two months. This means our allowed down time for 2009 has been mostly spent. In March and April the power grid and the Internet connection was working well which meant that our up time during those two months was excellent.

In February I replaced the ISA Server remaining the one Windows-based part of the system with a FreeBSD based solution. The ISA performed well but ran on a PC, which is not a recommended if you need to have it one 24/7 for years. The new solution used an embedded PC from Alix that is built with no movable parts and created to be an always on routing solution. It has been working flawlessly since the day it was installed.

Posted: 2009-05-10 by Erik Zalitis
Changed: 2009-05-10 by Erik Zalitis

News archive