Zero Downtime

Zero Downtime 

"In a 1998 report from IBM Global Services, unavailable systems are estimated to have cost American businesses $4.54 billion in 1996, due to lost productivity and revenues." 

Any businesses that relies on a system will have to anticipate that it will be unavailable for a while. However, how good is the system depends on the time of unavailability, how fast can it recover from it. What are the technical knowledge and other resources required to get the system available again.

For Michaelsoft DDS, the diskless network solution, have prepared few steps to prevent the system from unavailability. We always advice customers to purchase two servers to prevent system unavailability. From our recommendation, for maximum performance, each server will have 6 hard disk, 3 for Linux server, 3 for client PC's OS and software applications. The combination of the hard disk can be a 2+2, 3+2, 4+2 depending on the business needs.

The first 3 hard disk totaling up to 1.5 TB, stores the files to run the Linux server operating system, the rest of it to support mass number of workstation's read/write, and also to store their personal files. The next 3 hard disk is to store the client workstation's OS and software applications, such as word processors, image editing, movie maker, Auto CAD, and many more.

In the event of virus attack

If the client workstation was attacked until the system freeze, just a quick restart will solve the problem, as the temporary storage of virus will be removed. Client workstation was configured to boot it's OS and applications from the server.

In the event of network virus attack

No matter how many times restarting on the client workstation, the virus from other workstation will be spread through the network. This happens because of the server is infected with virus. The solution is to find out which server has receives updates and installation recently, and rollback to it's previous restore point.

In the event where applications is faulty or corrupted

If the applications failed to boot from the server, the installation of the application has encounter a problem, therefore it has to be re-installed or rollback.

To re-install, one of the workstation will be logged in as 'super workstation' where administrative work can be done, to re-install the application. This process involves logging into one of the client workstation, transfer the installation files from another computer through network sharing (plugging USB into the super work station will be exposed to the possibility of virus infection) and complete the installation of the application. Once the application has been installed onto the server, other client workstation will be able to run application. Or else, rollback to it's previous restore point.

Instead of the conventional way, cloning the installed files through network takes a lot of time. Diskless network solution is able to deploy the application in a single installation process.

In the event of re-installing the whole server

In the event where the hard disk in the server was found faulty, among the 3 hard disk, the 1st will usually be the most heavy duty, and it's the highest chance of being faulty, due to it's usage for booting the server and client workstations. This hard disk will be replaced, and the installation of the server has to be re-done again. The installation of the Linux server takes about 5 minutes using either the optical disk, or USB to install. To get the server up and available, the configuration takes about another 15 minutes, to get the client workstation's IP address booted from the server. While doing the installation, the other server which is still running will keep the system alive and provide as a host to duplicate the files to the new server. 

Conclusion

In a business, a downtime means a loss. By having at least two server, the system can achieve zero downtime.




Reference
http://www.dis.uniroma1.it/~irl/docs/availabilitytutorial.pdf