High Availability for SMB Sector - The Goldilocks Sizing Principle

So I am a FIRM believer that the SMB sector needs every bit of the same functionality and flexibility in their IT organization as the BIG Corportations. But I also am a realist that not everyone can afford all the bells and whistles that they would like to have. IT magazines have been pushing High Availability (HA) for years typically calling for Five 9's of availability. By definition this means your systems are available 99.999% of the time requested. I believe that the TRUE meaning of 5 9's gets lost in the equation. Generally speaking, from a Software and Infrastructure COST perspective the Higher the Availability requirement, the more expensive the solution. Therefore it is important that Business Leaders and IT Management work together to determine the TRUE requirement for your specific business application and not just going with the industry trend.

Example: If you run a typical business you want your systems UP and operational from 6 AM (for the early risers) to 6 PM or about 12 hours a day. Multiplied by 5 days a week and 52 weeks a year the total number of hours works out to about 3120 hours annually. So you want the elusive Five 9's of uptime - that equates to under 2 minutes of Planned and Unplanned downtime annually. How about Four 9's (99.99% uptime)? That works out to under 20 minutes of Planned and Unplanned downtime annually. Three 9's (99.9% uptime) works out to just under 190 minutes downtime annually - that is still just over 3 hours for the whole year. Still sound pretty aggressive? Two 9's (99% uptime) comes out at just over 31 hours downtime annually or 0.6 hours per week. To round out the computations 97% uptime is under 94 hours of downtime a year and 95% uptime comes out to 156 hours downtime annually.

To achieve the Two 9's of High Availability you may not need excessive Infrastructure but you would certainly need a high degree of controls about changes to your Production Infrastructure environment. A vast majority of outages are actually caused by human error, either thru the incorrect implementation of something or the incorrect testing of an update before placing it into Production. Anything above this level will certainly require your IT hardware being in what are called HA clusters - redundant hardware used to insure if server A fails that server B is around to pick up the load. There are Active-Passive Clusters where the second node is brought online in minutes if the first node fails as well as Active-Active Clusters in which the second node is up, online and monitoring the first node and can take over the required processing in seconds. Additionally some database and application software has the ability, for additional licensing costs of course, to work across mutliple cluster nodes, etc. As you get even more deep into the Fives you will find N+1 server architecture that is load balanced by redundant traffic cops - all with increased costs to your budget.

Now sometimes one or more of your CRITICAL applications needs this type of High Availability and that is fine. I worked for many years in 24x7 manufacturing facilities that REQUIRED this type of HA or else you may end up stopping the production line...and you NEVER stopped the production line. Similarly in the 1990s I worked for AT&T Bell Labs and any outtage of the Long Distance Network was inconceivable. The Online Banking applications commonplace today are typically built for Four or Five 9's. Common place today, your firm may have an online store like Amazon.com or similar and may WANT to be operational 24x7 - the point is that every application, every system has downtime and needs to have a maintenance window to apply updated, security patches and the like. Your online store revenue and ROI may not be able to support the requested Availability from the business.

If you and your business needs assistance in determining the right level of availability for your critical applications we can help. Simply use our Contact form and we can schedule a FREE initial consultation. Our firm employees experts in IT Infrastructure and we look forward to working with you to solve these issues in a cost effective manner for your business.

Rob Vorbroker
#li