Battery Backup

We've recently had a rash of UPS problems that confused us for a while. We have frequent blackouts or brownouts (more than I noticed before I got into IT) and it has caused some interesting moments. Usually a UPS will allow us to go through these short energy spikes without issue. Since our phone systems and data systems obviously need energy, and the world stops when communications stop, we try to keep these systems running for 15 to 30 minutes without power supplied from the outside world.

After a few recent events, we've found that our mega UPS systems don't stay up for more than 10 minutes and will not return to normal operations after power returns. Shortly thereafter, a strong burning smell permeates the data closet. Under closer inspection we were alarmed to find that one or more battery packs are very hot to the touch! As we removed the battery packs we found that they have begun to expand and what seems to be a chemical reaction is occurring where they leak.

As we've removed these batteries and replaced them the system will go back to functioning normally. So what causes this? Apparently after 3 - 5 years of use UPS batteries are prone to losing their ability to hold a charge. Once they do not hold a charge the UPS continues to try to charge them up, causing a very dangerous situation to occur inside the battery. Some reports have been given where they actually start fires.

So what can you do to monitor or prevent this? It seems that there is little to be done other than monitor and replace your UPS systems that are within this time range. We've evaluating the prospect of putting in thermal sensors to alert us to a battery pack that is getting hot. It amazes me that the one system you purchase to protect your electronic investments can ultimately provide their demise.
(Update 16May2011: We have measured the tempurature on some of the failing batteries (like today's) at over 160 degrees. Please use caution when removing these battery packs as they weigh over 40 pounds, are very hot and most likely are leaking acidic liquid when in this state.)

VDI - Storage

One of the two main hardware considerations for any virtual desktop deployment is storage. While the virtual instances will need computing power on the server side, you'll also need to consider storage capacity for each instance running. Many popular virtual desktop packages prefer a SAN connected to the system rather than a NAS. What's the difference? Glad you asked! The basic difference between the two is a SAN (Storage Area Network) connects and appears to the server as local storage. Whether it is through iSCSI connections or fiber channel, the server acts and communicates as if the storage that is carved out is a physical disk or array. The SAN storage space can be expanded on the SAN controller. Speeds are generally better for a SAN than a NAS. Another word you'll want to research is IOPS. Go to a tech net article here on calculating iops or (input/output operations per second). The higher the IOPS the better the system.

A NAS is a Network Attached Storage device. It is a lot of hard drive space that you can utilize over the network. Obviously it will not have as good of read/write speed as a NAS, but can be much cheaper to purchase and support. Fiber channel switches are not cheap and iSCSI connections can be real fun if you don't know what you are doing. The VDI system we are using prefers a NAS. However, if it doesn't work out, the other systems we'd use prefer connecting to a SAN for performance.

So what to do? We went with NetApp precisely for this reason. NetApp can run either as a SAN or a NAS on the same box. So, while we might use the NAS side for virtual desktops, we can connect to the SAN side for other storage needs! We'll let you know how the installation and training goes and how difficult it is to manage. You can get an idea on NetApp's site for free as a guest user. This is only for a limited time.