In the internet era, something profound has happened to traditional connectivity provider thinking and spending related to network service availability. Something equally important has happened to consumer expectations about service or app availability, as well.
Service and app providers using the internet as the access platform generally cannot expect to reach 99.999 percent availability. Neither can mobile services of any type. Virtually all fixed network services except perhaps voice also now will fail to reach 99.999 percent availability.
The reason is simply the math. Availability of any single device, in principle, could reach 99.999 percent availability, though that is not true of routers or servers. End user devices also never reach 99.999 percent availability.
But system availability is the sum of every single device in the delivery chain. And even if every device and network element had 99.999 percent availability (“five nines”), reliability shrinks with the number of devices.
It is demonstrably not possible for most end user internet experiences or applications, almost certainly. In the transport business, 99.999 percent availability is probably possible only on the portions of the network with the fewest total network elements, end to end.
As a rule, I think we safely can assume that any internet-delivered will not ever reach 99.999 percent availability at the user level. It also is safe to assume that mobile voice, text messaging and internet access virtually never reach 99.999 percent. Possibly only legacy fixed network voice can hope to do so.
It goes without saying that the Internet is a complex system, with lots of servers, transmission paths, networks, devices and software all working together to create a complete value chain.
And since the availability of any complex system is the combined performance of all cumulative potential element failures, it should not come as a surprise that a complete end-to-end user experience is not “five nines.”
Any end-to-end network, and the applications delivered over that network, have many single points of failure. Even if most network element has five nines levels of availability, and there are just eight total elements in series, with one device having 85 percent availability, and one device has 95 percent availability, total end-to-end system availability is just 59.87 percent.
That is calculated as 85%*90%*99.9%*98%*85%*99%*99.99%*95%, or 59.87 percent.
Redundancy is the way performance typically is enhanced at a data center, by a transmission network or application. But the total number of network elements for any single app also is going to be more than eight.
So one way of thinking about reliability or availability is that modern application delivery systems cannot actually meet the old “five nines” standard, end to end. There are simply too many network elements involved in any end-to-end system.
The business implication is that “five nines” increasingly is not possible.
But consumer expectations also are involved. Consumers know mobile phones are not as reliable as fixed network phone service. They are willing to make the trade off. They know computers, operating systems and web apps are not “five nines” available. They know they will have to reboot on occasion.
The business implication is that consumers can live with “less than 99.999 percent availability if value is high enough. The corollary is that they also might not be willing to pay what it theoretically costs to move closer to five nines availability.
The bottom line is that five nines cannot be provided, nor will consumers pay for it. Some level of availability less than 99.999 percent is acceptable and even expected. The question is how much availability is required to keep a user satisfied, as they might not be willing to pay much more--if anything more--to get that level of availability.
No comments:
Post a Comment