Linode Rocks (and Thanks for All The Fish)

This year has really been hectic thus lack of posts here. This will be a short post as well as I aimed to finish my tax matter this weekend (for Australia, personal tax return is due at end of October each year). It basically details some of the changes I had with my hosting arrangements for the past few months.

I migrated HostingFu from SliceHost to VPSLink last November, however HostingFu has actually been running on a Linode (aff link) in Dallas for a few months now, after the sale of VPSLink/Spry to Endurance International. I previously had pretty good review on their Xen VPS, and had 2 VPS with them — one OpenVZ and one Xen. Cameron and Dan have been great to work with, and they ran a great setup in Seattle.

However the sale ruined it. Migration of all servers to Boston is not well executed. After all I do not think I can recommend them any more. I used to have $1000+ of referral credit under my account but it disappeared into nowhere after the transition. My 8-core Link-4 Xen VPS now only can access one CPU core. So I decided not to renew, and a sample of their dying forums is also full of similar stories of people migrating elsewhere. It’s sad to see how a once great company has fallen.

Hello Linode

Well. I am no stranger to Linode — I became a customer when I first reviewed them almost 3 years ago, and had hosted my biggest site with them — until this week but more about that later. So when I decided to move HostingFu away from VPSLink, and was too busy to find a new host, I turned to Linode. I used that opportunity to write an article on how to deploy a Linode with their API, and that newly deployed node ended up running HostingFu.

It had been smooth sailing since May (as expected), and Linode even gave a free RAM upgrade back in June (a bit unexpected). $19.95/month for a Xen VPS running Ubuntu 10.04 with 512MB of memory, 16GB of disk space and 200GB/month of data — not the cheapest but probably no competitor in stability and flexibility at this price point. It’s now also running all my git repositories and a private Trac instance. Two thumbs up.

I also have a pair of Linode in Fremont running a busy Drupal site — and the web node (Linode720, Debian 5) is having uptime of 319 days. On September 12 my database node in Fremont went down (Linode1024, Ubuntu 10.04). By the time I noticed my Pingdom alert, Danny Ariti from Linode (their Australian support guy) has already created a support ticket and kept me informed throughout the 1 hour down time. Top quality support I’ll say. Linode rocks.

Bye Bye Linode, and Hello Crucial Paradigm

Then last week I made an important decision to move my busy Drupal site from the pair of Linode in Fremont to Crucial Paradigm in Sydney Australia. I have been thinking of moving that site for quite a while because it does not make sense not to, as 95% of visitors are coming from Australia. Why then should I deprive them with 200ms of latency across the Pacific Ocean? I went with Crucial Paradigm because I have reviewed them before — again due to my lack of time to do more research so I have to trust whom I have previously worked with.

There were some big disadvantages moving from Linode to Crucial Paradigm though. First of all it’s the cost. A pair of Linode1024 costs USD$80/month with 800GB of data, and that’s $81.63 AUD at the moment. On the other hand I am paying a lot more — AUD$400/month for one single VPS in Australia. It has better spec (6x E5620 cores, 6GB memory, 300GB storage — way more than I need), but lacking in data (600GB/month, after Aaron W generously doubled that for my package).

While Crucial Paradigm is also a great company, it still does not have the flexibility Linode gives me. Because my requirement of Ubuntu 10.04 they have to put me on a Citrix XenServer node (rather than HyperVM node). Which means rebooting need tickets, and I don’t get OOB console either. Obviously I do not have the flexibility to quickly deploy a new node to test something and decommission it (and costs me only the time it’s up). Private IP? API access? Rescue mode? Backup service?

Well, I’ll see how it goes. I am hoping a better experience for my visitors would offset the deficiencies of moving away from Linode. Hopefully.