What do you want, a pat in the back? It's awesome that you and your company are doing all those things and run extremely well, and the article is very interesting as a specific tale in the evolution of a particular service.

But then you start projecting all that into the rest of the world, with what appears to be very little actual insight into the rest of the world; showing a blanket dismissal of anyone who can't do the things you do; and refusing to acknowledge that many companies and products operate under very different parameters and priorities.

Thanks for the info and the data, and feel free to understand why I am ignoring the rest of the stuff.

I hate to burst your bubble - I work for a Fortune 15 company and it takes us around 4 weeks to quote, purchase, and source a single new server. During the flooding in Thailand last winter which impacted hard drive inventories at our major server vendor, HP, it took us about 8-12 weeks to do the same. If you want a cluster or more than one server, we go through an architecture review process that might add another 4 weeks to the process.

Granted, if you just want a VM, and we have excess capacity, we can turn that around in about a week, but you have to understand, large companies have procedures that you probably never even thought of:

A new server has to be spec'd out; the requirements have to be gathered from the business owner of the server (OS version, CPU, memory, disk space, network and SAN connectivity). This is all put into a worksheet that is stored in a document management system. Quotes are obtained from a vendor. A design review meeting takes place to make sure no parts are missing from the quote (9 times out of 10 we need to get corrections on the quote because something as simple as a cable, KVM dongle, or spare power supply is missing). Now we go the the purchase approval process - much paperwork is filled out. Depending on the $ amount, between 10 and 20 people might need to electronically approve of the purchase. At any given time 10-20% of those approvers might be on PTO and you either have to wait for them to get back or contact their admin. assistant to approve for them.

With that all done and the purchase approved, now the PO is cut, wait 2-4 weeks for hardware to arrive (8-12 weeks if inventory is constrained due to circumstances out of your control like the Thailand flooding). More paperwork is done - the system is physically racked and an OS is loaded. There is a database where the server needs to get loaded into. Security hardening takes place, Nessus scans, and we ensure it is on the patch management schedule to receive patches. Configuration is done, and if all went well, we're now 4-6 weeks into the process and the server has been turned over to the application owner to install their application. After they install their application we take it back for a day or two to configure and test backups for them.

Your reply contains a great amount of naivety about the way most large companies work. What works for you, where you are developer and operations all in one person, doesn't work when you scale it out and have to work across many departments. We have Linux, Windows, VMware, Storage, Network, Security, Database Admin, and Application Support/Operations teams that all might have to touch a box before it is ready for the application to go live.

By the way, we're in the infancy of looking at cloud, but chances are we will have a private cloud long before we trust a public one - we can't risk our data leaving the data center.