The LyonsHost migration is going. So sorry that I haven’t been writing as much recently (OK, actually I’ve been writing my hands to the point of cramping, but not in this space.) We’ve been very busy. Personally, I’ve been recovering from knee surgery, and the LyonsHost transition I wrote about has been slower than expected.
So, while everyone else is facing Snowzilla, we are pushing forward on the last leg of our journey to improved products and service. We moved two thirds of our servers successfully. We had a rough start, but the second phase went much better. Over the next few hours, we are moving all but one remaining server. If everything else goes faster than we anticipate, we have the option to complete the work tonight, otherwise we will close out the move tomorrow night, Sunday January 24.
What should you expect? If you haven’t heard from us about tonight’s move, you are already on one of our new servers! Hold tight, you will have a small interruption on Sunday, when we move our Domain Name Server (DNS), because the DNS tells the Internet where every website is located in our network by its IP address. We will have to shut the DNS down briefly and restart that server in its new home, thirty to sixty (30-60) minutes later. Depending on which server your sites reside, you may not even see us blink because of Time to Live (TTL) which tells the Internet how often to check our DNS for changes. If your provider checks DNS just before the change, TTL could mean that your site will not show any changes from where you sit. It’s a nifty trick. Of course, if your Internet provider checks our DNS during the move, you will be unable to access your sites for the entire procedure. I’m sorry, there is no way that you or we can influence that timing.
Do the new servers make any real difference? You bet. With one exception over the past couple weeks, everything on the new platform has been very stable. The servers are noticeably faster. The servers are showing fewer signs of strain in terms of load, memory use and HD access. In many cases, the load indications are 1/4 or less than our previous system.
The anomaly we experienced on Thursday was a telling example. Unlike past episodes, the problem contained itself to a single server. We contained the problem and ultimately tracked it down to software that keeps statistical data for our mail servers. That made it look like the problem was happening in whichever user account the software was examining at the time. It was frustrating, but after being fixed, every measurement shows a strong and stable system.
We have more good things to share coming up. Thank you for your hosting and email business. Oh, the knee, I’m already walking around and cycling on my indoor trainer. It’s going to be a great year.
UPDATE: There was a problem with cPanel licenses which prevented us from performing quality assurance (QA). We will pick this up again tomorrow.