What are the repercussions of a Software as a Service outage? Obviously, there is significant monetary value being lost during these periods. Customers may not make purchases on eCommerce sites or could simply take their business to a competitor. The basic sources of downtime are attack, overload, no power, bugs, and breakdowns. They each have countermeasures which should mitigate most effects. Attacks can be avoided using security tools such as firewalls. Overload can be anticipated using monitoring tools, scalable architectures, and load testing. The absence of power can be avoided by using diesel generators. Bugs can be healed by frequent patching. Breakdowns can be repaired using redundancy and backup. The main reason why many of these problems still exist throughout the internet is that many users do not employ countermeasures.
The number of systems accessing the internet without protection is steadily decreasing. This may or may not be due to increased publicity of disaster scenarios. However, it could also be due to increased instances of additional inherent protection. This can include automatic updates, defaults set to secure, or better software in the marketplace. This should not lull the user into a false sense of security. It is an important part of any system manager’s job to plan for one of these contingencies. Many system managers already have extensive virus and malware protections. Today, we are trained from day one to avoid exploits and create our programs with more inherent security. However, this is not always the case with scalability.
Any program built to live upon a server should be designed for scalability. At some point, the service will face either high I/O or computational loads. It is important that the program handles that load in an effective manner. Once the system is in place, it is extremely expensive to rewrite or replace it. Additionally, popular services will continue to grow and test the system. If the scaling is not performed fast enough, then many users will abandon your service all together.
When there is an outage of a SaaS, there are no if, ands, or buts. The software product is simply gone. Any data or files are inaccessible to the customer. This puts stress upon the SaaS provider to maintain 100% uptime. If 100% uptime cannot be maintained, then the service disturbance should be localized to a small number of users or feature. Localized disturbances should provide alternate venues through which users can access their data. Any disturbances have a critical period through which they must be fixed. For a business operating 24/7, there should be no downtime.