What Downtime Is Acceptable for Your Business Applications?
How much downtime is too much? Required levels of availability for businesses are steadily rising. Let's talk about how to make sure your users won't run into issues.
Join the DZone community and get the full member experience.Join For Free
There was a time when 99% availability was considered good. But that meant about 4 days a year of unplanned downtime. Slowly but surely, the required levels of availability increased to less than a day a year (99.9%), less than 20 mins a year (99.99%) and now to 20 seconds a year (99.999%)
But is that good enough? Often the answer is no.
The issue is not just how long in total a system is unavailable, but exactly when a system is unavailable and what that unavailability means to the systems it connects with. When each of the individual systems that make up a complex business application is down, then the whole business process may slow down or stop.
If you have 30 systems connected together (which is a lot less than many complex business applications require) to make a business process, and each one may be down for just a half a second a week, that can effectively mean that you can have up to 900 (302) events in a week when a business process could fail. So just because each system is performing at 5-nines (99.999%) you may still have an issue.
With all the redundant systems, and process in place to maximize uptime, there is still always some systems that will inevitably fail, and even with the most advanced technology jumping in to automate the swapping-in of replacement reductant systems things can slow down to the point where business processes will be affected.
Resilience to even small interruptions can be improved by using technology to queue requests between systems and retry if a request fails, so instead of a total failure, you will just see a slowdown in the processing of a request. This is precisely what middleware does for many business applications.
The challenge then becomes one of command and control.
You need to have a central view of the availability and performance of each and every system in your application stack, including the middleware. And you need to know how each complete (and incomplete) transaction is performing across the entire ecosystem.
You should be able to identify potential slowdowns that could cause agreed service levels to be missed before they are missed.
Nastel AutoPilot is the solution that provides the level of monitoring and transaction resolution that is high enough to ensure availability to the levels that large-scale modern business applications demand.
Unless you have total visibility of each business transaction, the detail on the performance at and through every system you will find yourself in lengthy, expensive debates as to what parts of the process are responsible for any deviation from the desired state.
Customers who use Nastel AutoPilot are able to maximize their customer experience while managing to their operations and applications development budgets.
To find out more, visit www.nastel.com.
Published at DZone with permission of David Liff, DZone MVB. See the original article here.
Opinions expressed by DZone contributors are their own.