You probably heard about the typo that stopped the internet – how a few weeks ago an Amazon Web Services error brought down a huge chunk of the Internet along with many key sites for hours. The incident brings up some interesting thoughts about hosting from a software perspective vs. hardware perspective.
As most people know, by using Amazon Web Services, many companies use the Cloud for storage. So instead of having one server with all their data and programs running on it, when they want to access something, they are technically calling API’s for a piece of software that is not on a desktop or office server, but in the cloud.
The cloud itself is not the reason for the vulnerability that brought the Internet down. They have the same or better redundancy as localized hosting services. Amazon has servers around the world and can even create pop -up servers to handle excess traffic (like for the Superbowl).
So why were things down for so long? It’s because of how the Internet handles routes. And that is by using major trunks, huge pipelines, where routers cache data along the way. This is also a “self-healing’ systems so that if one router is out, data flows to another one.
What happened in this case was that a developer at a data center in Northern Virginia was debugging an issue and executed a command intended to remove a small number of servers. He accidentally typed in a command which removed more servers than intended. And while it was only moments before the error was discovered and actual total server outage lasted only 15 minutes, because so many servers were out, it took 3 hours for the system to flush the cache of bad data through.
So, the problem was not a lack of redundancy in the software-as-a-service (SaaS) model, the problem was that Amazon did not have checks in place to identify when a particular (software) command doesn’t make sense.
Enter a new model, hailed by some companies as a preferable alternative: Infrastructure-as-a-service (IaaS).
Companies like Expedient, headquartered in Pittsburgh with locations along the East Coast, offer a new take on the old-school model of IT, one where companies can still have dedicated, redundant hardware of their own, but instead of purchasing it and keeping it onsite in their offices, they rent the hardware from a provider like Expedient, who maintains banks of physical machines for their clients.
Expedient and similar companies are true IT companies. Clients engage them to set up infrastructure, servers, networking bridges, and firewalls. Thanks to back up generators and redundant systems, Expedient promises 100% uptime, with hot back-ups triggered by mere seconds or milliseconds of noncommunication to a machine, getting clients up and running within less than a minute.
For those companies who wish to protect themselves from the kind of vulnerability created by the Cloud, Expedient offers a more localized and hands-on IT approach with many of the business advantages of the cloud model – no capital expenses or depreciation, no need for IT staff to support and maintain machines, but rather an ongoing, affordable operating cost. In a sense, it’s the best of both worlds for these companies: they can still have and manage their own servers and enjoy more localized access, without having to buy the hardware or hire the staff to support it.
So, what’s best for my company?
Choosing the best model for your company depends on what you’re looking to do, and whether you want to have IT control over it. Do you like the idea of having your own servers, that you can log in to any time? Do you have your own IT staff, and just don’t like the idea of investing a lot of money in server hardware? Or are you without IT staff and uncomfortable with the Cloud, in which case you can work with a company like Expedient. If so, the newest old model might be for you.
Bottom line, companies have some great choices these days. The Cloud offers tremendous advantages with SaaS, and now IaaS offers back more server control to companies who need or want it.