There are no good web hosts, period. Now I am sure this will piss a ton of people off, but please let me refine this. Unless you have a $10,000 a month or more of a hosting budget, you are shunned, given sub-standard equipment or even worse - hung up on four times in one day for a support call. Yup - that was my day today with the latest provider I chose, theplanet.com. Now as most of you know the Planet is a super-large hosting facility located in Texas and they run almost all of the mid-level hosting companies back end operations. They have a great data center, great machines, and great support - according to their web site. Let me, my friends, fill you in on my experience over the past few days, especially today.
About 4 days ago, one of our smaller sites was victim of a Hack. A nefarious individual posted a eBay phishing site and then spammed upwards of 5000 people to look at it and enter their eBay information. First off, very bad, second - even worse when it is on your server and you get a phone call at 6am from eBay security. We killed the hack and fortified the server in less then 30 minutes after the call, then spent the day making sure things are good.
A few days pass, and we come to today. I woke up to check my email only to find that nothing was responding. A four core, dual processor machine with dual 1000Mb up links the the net, sitting on a Tier 4 backbone and nothing. Not even a ping. I quickly called support and after 27 minutes on hold (yes, 27 minutes) I got a newbie that said submit a reboot request. Then I got one ping. I told him and then I was hung up on (possibly being transfered, so a little pissed). Called again only to sit on the phone, go through 3 people and get hung up on once again. After dealing with this for about 2 hours, I receive an email from The Planet that stated, we are looking into 74.5Mb/sec traffic to our server as being a possible DDoS attack. I was pissed. I attempted calling them to speak to the security area that sent the email, but apparently they don’t have phones there. I left countless messages in the tickets and emailed them for over 2 more hours. A second email then shows up stating that yes, in fact it was an attack and they were putting me behind their Flood Guard protection. When I heard this I was very happy, I thought, great - finally someone is helping me out here. I come to find out the server is still totally unresponsive so I update the ticket once again - “Flood Guard is not working, please block all traffic to your network and my IP” - an easy request.
About an hour later I get an update to the support ticket stating that the attack was from one IP (with the address) and with the port the attack was coming in on. I thought great, if I can SSH into the machine I would be saved - only the machine was dead… nothing worked. I updated the ticket for them to go in through the console.
To make this a bit shorter and to stray away from the profanity welling up in my soul, let me just say after 10 hours, the server was back up. Unsatisfactory. Unacceptable. Period.
When I started to look at the traffic patterns and numbers being reported by the server - supposedly with dual NIC’s on huge pipes… was a little off-putting. Why was the attack limited to 74Mb (mega-bit). The connection was a 1000Mb connection, so surely there was adequate space to get through - even after I see the CPU usage was at 47% (you have to split that as well - dual / dual core processors). Also, why wouldn’t the management port respond - it is supposedly on another NIC all together? Well the answer looks like an easy one, but I haven’t confirmed it yet. The servers at the Planet (dedicated ones, expensive ones) are either throttled at the router level or they do not install 1000Mb cards when requested, they just use 100Mb. I will look into further and update, because if it is true it would be sad.
Now I have used some of the best out there, and out of all of them RackSpace is the best - hands down. They take support to an insane level - but in no way can I afford them… or so I thought. In my research to build the hosting infrastructure needed for my new project ContentMotor, I came across a little company called Mosso (owned and created by RackSpace!). It is a true Cluster / Grid environment that you can buy instances on for $100 month and it gets you a lot 80GB of Storage and 2000GB of bandwidth. We have been in discussion with them for a bit and have selected them to host the accounts needed for our systems, but I thought why not use them for my sites as well and just weigh overages compared to these massive dedicated servers? What did I find?… it is exactly what they say - “See Why the System Beats the Server”.
The sign up, set up and use thus far has been stellar. The support - ON A SUNDAY, was available with no wait times, live chat and email and I have had two follow up calls today to see if everything is okay. These calls came from no tickets, no emails, no requests - they were actual service calls to check on an actual customer. Wow. I am sold. I am going to give it a few more days to make sure there are no DNS issues and post up some speed stats and latency stats but I have to say I think I found my web host home…
For more information check out Mosso.
That is an ambitious title, but you haven’t seen the product. If you have, you know I am right. Earlier I covered Amazon S3 as a revolutionary service - but what is so revolutionary is the software being produced to use it. That leads me to Bandwagon - a auxilliary program for iTunes that uses S3 to back up all your music from your iTunes library. Safe. Secure. Forever. Amazing.