There is a paradigm shift that all of us network engineers have seen for a while now. When I started in networking, you could barely find a server that could match your LAN’s data rate (100Mb), the best services you could find were in the enterprise, and data wasn’t expected to have the same up-time as your phone system. That has obviously long since changed, and perception has with it. In the enterprise environment, your top competitors are dropbox and gmail. Not that you are actually competing against them, but your users are well versed in them, and they can’t understand why their e-mail and file sharing don’t work like these free applications that never go down.
Of course google or AWS (what dropbox is built on) has network failures like anyone else, but the applications never go down. No matter how much engineering or infrastructure, at some point there will be an outage. You can see that in the SLA’s that cloud providers put out, Gartner summed it up nice:
However, AWS and HP both define their SLA not in terms of instance availability, or even AZ availability, but in terms of region availability. In the AWS case, a region is considered unavailable if you’re running instances in at least two AZs within that region, and in both of those AZs, your instances have no external network connectivity and you can’t launch instances in that AZ that do; this is metered in five-minute intervals.
So what amazon and HP (and every other cloud provider is telling you) is that you can’t build networks that don’t fail. So why build any network for 5,6, or “X” 9’s, instead require your applications to be built around the network requirements, not the other way around, which has been the status quo for as long as I have been in the business. Now I know that the network engineer doesn’t just get to turn around one day and demand “all applications be built around the network requirements,” but it is worth talking to the application developers and communicating the new requirements now, rather than later. Also worth taking into consideration when considering moving applications to the cloud. Could you in fact build a cheaper private cloud to your specific requirements and locations if you didn’t have to build for such high redundancy?
0 Comments.