I need help calculating the overall availibility of the following setup; In this

Question

I need help calculating the overall availibility of the following setup;

In this week's Application, you extend your previous work, as illustrated in the diagram below. The architecture has been slightly modified to include a dual LAN (local area network) that connects the users, both local and those accessing the system via the Internet. Once again, calculate the overall availability of the system, but this time, account for the failover characteristics of the components that are provided in the reliability diagram below.

In addition to the reliability diagram, consider the following characteristics of the system:

The dual LANs are load-shared. That is, during normal operation, the user traffic is split between the LANs, and both are active.

Should a LAN fail, traffic is simply rerouted over the surviving LAN. Thus, there is no failover time, and there is no chance of a failover fault because both LANs are consistently monitored as part of their load-sharing operation.

The mean time to repair (mtr) a Linux node is 4 hours.

Should the primary server crash, the cluster takes an average of 10 minutes to fail over. It will fail over properly 98% of the time.

Two percent of the time, the system will be unable to fail over. In this case, the primary data center will be down, and operations will switch to the Chicago backup data center.

If operations have to be switched to the Chicago center, it will take an average of 2 hours to bring the Chicago data center up. Failing over to Chicago is unsuccessful 5% of the time.

Thanks guys, I'm totally confused on this subject.

Explanation / Answer

Hi,

I will explain both the worst case and best case scenario of this configuration mentioned above.

Normal operation

CASE 1:In case of normal operation where all the nodes are working properly the system is available i.e a=.997.

In this case the user connects to the LAN and availability of LAN a=.9999 now even if one of the LAN fails there is a switch over to other LAN and there is no delay associated with this failover and loadsharing is performed.Now if the user has to access the Linux system through the network(LAN) and the availiablity status of this machine is a=.997 (note we are considering that all nodes are working fine in this case) now the server(linux) is connected to backend storage and this storage is available a=.9995 therefore considering this situation the overall system in case of normal operation is said to be available a=.997 times.

Failover scenario:

CASE 2:Now assuming if one of the nodes(linux server) fails in the overall configuration user is able to access the LAN and its available a=.9999 now as they go through the LAN they reach the linux server and if one of the node fails the time taken for switch over to other node is 10 minutes now the availablity reduces by 10 minutes,since the probability of failure of this node is 2% the availabilty at this stage considering fail over is reduced by 10 minutes adding 10.997 minutes latency to the overall performance of the system.At this stage 98% of the time the system is available up and running in 10.997 minutes.

CASE 2:Now if you consider the same failover with one of the node Linux fails and the failover adds 10minutes latency to the overall availabilty of the system adding 10.997 minutes latency.However if we consider the worst case scenario in which the failover has not happened to the other node (other linux) machine.which is just 2% times.Now in this case it first adds 10.997 minutes latency in this case the entire primary datacenter in this case newyork datacenter is down there fore we have to switch to the secondary datacenter which is chicago data center in this case.Now failover to this chicago datacenter itself will take up to 2 hours thus delaying the overall operation by 2 hours 10 minutes and 0.997 seconds.Thus the system is available almost 95% of the time with 2 hours and 10 minutes and 0.997 seconds latency.Now if the system is not able to backup in case of the 2 hours delay also the which is 5% of the time.Now to acces the server on the Secondary datacenter will take 2 hours 10 minutes and .997 seconds .Now once the system comes up .i.eLinux server on new york datacenter comes up after 4 hours the average latency is 4 hours minus the 2 hours and 10 minutes which is 1 hour 50 minutes.

Hope this helps...HAPPY ANSWERING!!!!!!!!!!!!!!!!!

Related Questions

Navigate

• Browse (All)
• Browse I
• Subjects

Integrity-first tutoring: explanations and feedback only — we do not complete graded work. Learn more.