RJM62
Touchdown! Greaser!
- Joined
- Jun 15, 2007
- Messages
- 13,157
- Location
- Upstate New York
- Display Name
Display name:
Geek on the Hill
About a year ago, I decided to purchase a VPS account on a cloud in my never-ending search for what I consider the Holy Grail of hosting: 100 percent uptime.
That didn't work out as planned.
In all fairness, the hosting was actually very good. But the uptime wasn't "100 percent" as promised. And I'm not talking about a few pings or a few minutes. Some of the outages lasted for hours. That's unacceptable to me.
But also in fairness, some of the outages really weren't the hosting company's fault. Two of them, which occurred within the same week about a month ago, seemed to have been caused by a problem with either PHP or Apache. On a hunch, I decided to update PHP and rebuild Apache, and the problem disappeared. Why? No idea. But although I agree that the problem wasn't really the hosting company's fault, I was mightily annoyed that it took more than two hours for the support techs to get the tickets.
Last night there was another problem. It was obviously a network issue. I was getting packet loss ranging from 25 to 60 percent. I submitted a ticket, and within 10 minutes of the support tech reading it, the problem was solved. Another machine in the DC had been zombied and was spewing forth UDP traffic. It was isolated and suspended, and the problem cleared up immediately. The problem, once again, was that it took well over two hours for the ticket to reach the tech. So the 10-minute fix took nearly three hours to execute.
I know that there's always a queue for support, but two hours seems unacceptable to me -- especially in view of the "100 percent uptime guarantee."
I asked to be compensated for the outages last month, and the hosting company refused, stating that the problem was not their fault, per se. (This was the Apache/PHP thing.) I countered that it really didn't matter whether it was their fault or not. What mattered was that it took upwards of two hours before the support tech even got the ticket, during which I couldn't access the server to try to fix it myself. I lost that argument.
So I started my quest again, and I found that VPS.net has had pretty consistently good reviews. I bookmarked their page and waited until the next problem, which happened last night.
So I had a little online chat session with Nick Nelson, the Managing Director of VPS.net, and decided to give them a try. I opened the account this morning, and within minutes of their receiving a copy of my driver's license (why they needed it, I have no idea), my account was activated.
Of course, all that meant was that I had an account and a number of "nodes" available for my use. I still had to build a VPS on them. That was easy enough: I selected a centOS 5.3 LAMP image, and it was up and running in minutes. Then it took maybe another 45 minutes or so to install cPanel, maybe half an hour to rebuild Apache with the modules I needed, about an hour to migrate the 30+ accounts over, about a minute to create the reverse proxy needed by one of the sites (the automated migration in cPanel missed the reverse proxy), five minutes to install and configure CSF, and about ten minutes to set up rsync for the optional remote backup.
The VPS works great so far. I still have the old VPS running, limited to pulling DNS duty until the IP changes propagate. But the page loads and script executions are noticeably faster on the new VPS, and the total resource loads are much lower.
But perhaps just as importantly, my two tickets to VPS.net support were responded to within minutes. They were simple, sure: One was a question about the CSF configuration, and the other was a request to set up rDNS. But it doesn't matter that the tickets were simple. The point is that the tickets reached the support engineers in minutes, not hours. Very nice. That's what I like to see.
So we'll see how this stage of my journey to find true "100 percent uptime" goes. I'm impressed with VPS.net so far, and the server is working beautifully; so I'm optimistic.
-Rich
That didn't work out as planned.
In all fairness, the hosting was actually very good. But the uptime wasn't "100 percent" as promised. And I'm not talking about a few pings or a few minutes. Some of the outages lasted for hours. That's unacceptable to me.
But also in fairness, some of the outages really weren't the hosting company's fault. Two of them, which occurred within the same week about a month ago, seemed to have been caused by a problem with either PHP or Apache. On a hunch, I decided to update PHP and rebuild Apache, and the problem disappeared. Why? No idea. But although I agree that the problem wasn't really the hosting company's fault, I was mightily annoyed that it took more than two hours for the support techs to get the tickets.
Last night there was another problem. It was obviously a network issue. I was getting packet loss ranging from 25 to 60 percent. I submitted a ticket, and within 10 minutes of the support tech reading it, the problem was solved. Another machine in the DC had been zombied and was spewing forth UDP traffic. It was isolated and suspended, and the problem cleared up immediately. The problem, once again, was that it took well over two hours for the ticket to reach the tech. So the 10-minute fix took nearly three hours to execute.
I know that there's always a queue for support, but two hours seems unacceptable to me -- especially in view of the "100 percent uptime guarantee."
I asked to be compensated for the outages last month, and the hosting company refused, stating that the problem was not their fault, per se. (This was the Apache/PHP thing.) I countered that it really didn't matter whether it was their fault or not. What mattered was that it took upwards of two hours before the support tech even got the ticket, during which I couldn't access the server to try to fix it myself. I lost that argument.
So I started my quest again, and I found that VPS.net has had pretty consistently good reviews. I bookmarked their page and waited until the next problem, which happened last night.
So I had a little online chat session with Nick Nelson, the Managing Director of VPS.net, and decided to give them a try. I opened the account this morning, and within minutes of their receiving a copy of my driver's license (why they needed it, I have no idea), my account was activated.
Of course, all that meant was that I had an account and a number of "nodes" available for my use. I still had to build a VPS on them. That was easy enough: I selected a centOS 5.3 LAMP image, and it was up and running in minutes. Then it took maybe another 45 minutes or so to install cPanel, maybe half an hour to rebuild Apache with the modules I needed, about an hour to migrate the 30+ accounts over, about a minute to create the reverse proxy needed by one of the sites (the automated migration in cPanel missed the reverse proxy), five minutes to install and configure CSF, and about ten minutes to set up rsync for the optional remote backup.
The VPS works great so far. I still have the old VPS running, limited to pulling DNS duty until the IP changes propagate. But the page loads and script executions are noticeably faster on the new VPS, and the total resource loads are much lower.
But perhaps just as importantly, my two tickets to VPS.net support were responded to within minutes. They were simple, sure: One was a question about the CSF configuration, and the other was a request to set up rDNS. But it doesn't matter that the tickets were simple. The point is that the tickets reached the support engineers in minutes, not hours. Very nice. That's what I like to see.
So we'll see how this stage of my journey to find true "100 percent uptime" goes. I'm impressed with VPS.net so far, and the server is working beautifully; so I'm optimistic.
-Rich
Last edited:
