• available 24×7, minimal hold time
• ability to setup SOP (standard operating procedure) of what needs to happen if anything goes down
• escalation procedure — try to restart service, clean this and if that doesn’t work only then call one of predefined phone numbers
• etc

There are monitoring tools that are available for EC2 specifically (see my previous post here) or just standard networking tools (Nagios, Spiceworks, etc) but most of them still require us to hire staff to react to unexpected events

I looked into GoGrid and while it has a potential, it is not suitable for us just yet.

Positives:

• 24×7 support — don’t know what it means. My expectation is outlined above
• Graphical interface — while I didn’t like the design, it was user friendly
• Free Load Balancer
• Free Public IPs
• Windows 2008 Images

Negatives:

• Cannot use my own image, cannot backup/clone an image that I setup within GoGrid — that pretty much explains why they don’t have auto scaling (next bullet point)
• No auto scaling (roadmap)
• Even if machine is “off” — they still charge you money as if it was ON. On their roadmap.
• Only 4GB machine gets its own single Xeon core (equivalent to P4 2.0 chip). Lower RAM images get fractions.
• No multi-core support
• While their Cloud Storage pricing is inline with Amazon, Data Transfer is $.50/GB Vs Amazon’s $.17/GB for first 10 and then goes down from there. Significant difference
• Amazon offers CDN and GoGrid doesn’t
• Amazon offers multiple datacenters and GoGrid doesn’t