This series focuses on how to make your company’s infrastructure more efficient, more secure, and more cost effective. In Part I we discussed why it is important to employ a hardware firewall and leave room for growth; here are steps 3 through 5:
3. Have more than one physical server? Use Virtualization.
If you are maintaining more than one physical server, consider running virtual servers. Virtualization is fairly easy to use, and usually inexpensive considering the ROI; in some cases it is even free.
Although the uses for and benefits of virtualization are vast and potentially complex, initial integration is probably not as difficult as you think. “Configure” your virtual server by telling the software how many CPUs, memory and disk space to use, point it to your installation CD, and start it up. Your virtual server will boot in a window- then you can install and configure the operating system just as you would on a physical server. You can even convert existing physical servers to virtual servers with just a few clicks.
Why is this worth your time? Cost and Convenience. A virtual machine gives you the ability to remotely reboot, shutdown, start and access your servers via a remote console – without the headache of buying and maintaining additional hardware (naturally, the “host” machine needs to be powerful enough to handle the load). You can even take “snapshots” of your virtual servers so that you can rollback to an earlier date if you mess up a configuration and can’t recover.
4. Build Redundancy into Your Infrastructure
The worst thing an IT department can do is have a single point of failure with no simple way to recover. If your physical server dies, you’re left in a difficult position, since most operating systems are tightly bound to the specific hardware on which they were installed. Luckily, there are ways to greatly reduce the chances of catastrophe. The least expensive and easiest technique is redundancy. To illustrate our point, let’s look at the most common failure occurrence of modern hardware: the disk crash (cue ominous music).
Modern operating systems and many servers already support RAID (Redundant Array of Inexpensive Disks) configurations. There are different flavors of RAID, but most of what you need to understand can be summed up with the two words “mirroring” and “striping.” “Mirroring” means that all data is written to 2 disks. If one drive fails, then the other automatically takes over. “Striping” is used to double disk performance with reads and writes alternating between 2 disks, doubling throughput. RAID 0 is striping and RAID 1 is mirroring. Beware: RAID 0 is not “Redundant.” If one disk is lost in a RAID 0 array, then you will lose all of your data. For the performance of RAID 0 with the redundancy of RAID 1, look towards RAID 10 or RAID 5- high performance with redundancy, but these configurations will require more disks to implement.
Most RAID software and modern server hardware supports “hot swapping.” If a disk fails, you just pull it out and plug in a new one. If you have extra disks, you can even provision one as a “hot spare” that will immediately take the place of a failed drive. The RAID software will rebuild the volume and you won’t even need to shutdown the server.
For slightly more cash you can get hardware RAID, which provides the added benefit of being OS independent (other than drivers), and offloads processing from the server’s CPU to the RAID card. These usually cost approximately $150-300. With 1 TB disks selling on Newegg for $60, it’s almost criminal not to use RAID, and the same technique can be used for other software and hardware. Always ask yourself; where are my single points of failure, and if something goes wrong what is the plan?
5. Change Management
Possibly the most important, and most overlooked discipline in IT, Change Management boils down to the following:
1. Keep it as simple as possible- but no simpler (KISS; Keep It Simple, Stupid).
2. If it ain’t broke, don’t fix it.
3. If you do fix it, make sure that you can “unfix” it.
4. Document, Document, Document!
Don’t add software you don’t need, if it’s not part of the requirements, and it accomplishes nothing – why add another complication and point of failure?
Always have a “back-out” or “rollback” plan that you can execute in case your changes don’t go as planned. A plan could be as easy as making sure that you have a snapshot of a virtual machine so that, in a pinch, you can restore it, or it could be as complicated as documenting and saving your entire Web server configuration, including software versions, all supporting configurations, permissions, credentials, etc. In other words, if everything goes South can you get things back to the way they were before you started?
In a well-documented infrastructure you will know what services could be effected by a proposed change. Before you make a change to something, consider who and what will be impacted.
Have any questions about our 5 Crucial Steps, or IT in general? Feel free to comment below!
Thought this article was interesting? To keep up-to-date on business and technology news follow us on twitter @NECS_LLC, or visit the NE Computer Solutions Facebook page!