“Hi, the main server is down.”

       Have you ever had one of your main servers fail?  Was it down for 4 hours?  A day?  Two days?  More?  What was the impact on your company?  No access to key documents?  Could not send or receive email?  Could not use your primary application?  During the period it was down, did you mentally calculate a cost for the lack of employee productivity and lack of customer responsiveness?

     Just like buying insurance, you can buy r e d u n d a n c y to help protect your business against many of the common causes of server and internet failure.  And with the price of hardware continuing to fall, it may not cost as much as you think.  This NextWave article presents a few common strategies that can be used to avoid extended periods of down-time.

     To start with, we need to analyze the major points of failure.  For a server, from the hardware side, that could be a disk drive, controller card, CPU, RAM, the power supply, the motherboard, or even a cable.  Servers can also crash from software problems such as corrupted files or viruses.  Another key point of failure is the network switch that has one cable plugged into it for each of your users.  For internet connectivity, failure can occur because of a problem with a router, the broadband modem, a physical line problem, or a problem at your internet service provider (ISP).

     To protect yourself from server hardware failure, Beachwood Systems always recommends use of mirrored (RAID) hard drives within a server so that if one drive fails, there is another to take over.  Many of the newer servers are also designed to accept a redundant power supply and fan kit.  Although this strategy protects you against the most common problems, it still does not protect you against many others.  For that we recommend (when possible) use of identical servers with one being used in a capacity that is not critical.  If a critical server has a hardware problem you simply move the hard drives over to the non-critical server and you’re back in business.

    To protect yourself from server software failure, there are two good strategies.  One is to setup a real-time backup of important data files such as company Word and Excel documents so that if a server goes down, people will be able to access a document from the alternate server.  It is, of course, imperative to have good tape backups each and every day.  The other strategy is to implement the capability to restore directly from tape, even if the operating system will not load.

     A more expensive approach to redundancy is to set up a server cluster.  Under this approach two or more servers running Windows2003 Server Advanced Edition are connected to a common external disk array.  Users connect to the cluster and either server can handle the processing, providing instantaneous switchover if one of the servers fail.  It’s really quite remarkable.

     Has your company become dependant on the internet for email, vendor web site access, etc.?  We suspect the answer is yes.  Broadband internet connectivity using T1, DSL, and cable is much better today than in the past.  And yet, ISP’s do still have problems.  One strategy for redundancy is to get a second internet connection from a completely separate source.  For example if your business can get both DSL and cable at its location, then get one of each.  There are hardware devices on the market such as Xincom’s XC-DPG602 Twin WAN Router that can accept two independent internet connections and allow you to use all of the band width both incoming (downstream) and outgoing (upstream).