Light on size, not features

The B12S may be small, but it offers a set of management tools that, if not best in class, challenges many competitors. In addition to the small, easy-to-navigate control panel, I had the option of a CLI, a browser-based GUI, and SANWatch, a Java-based application.

All the management interfaces are functionally equivalent, so after setting an IP address consistent with my network using the control panel, I was able to switch between the browser and SANWatch with ease to manage the array. Thanks to the GUIs' intuitive designs, I had no trouble provisioning storage, monitoring the status of the hardware, or keeping an eye on the workload. One major difference between the two: The browser version lacked the rich and detailed online help I found on SANWatch.

I liked that you can create separate passwords for read-only access, maintenance tasks, and configuration, but other aspects of the security tools need improving. For example, I was able to change a password without having to type the previous password, which means that anyone passing by an unattended console could do the same. Despite these few glitches, the B12S's management tools are capable and easy to use.

The B12S maintains a detailed event log that traps error conditions and triggers the onboard alarm with custom thresholds. In addition, you can send critical warnings to an admin using messaging systems such as e-mail or SNMP.

Although not the most polished I have seen, the management tools of the EonStor B12S are adequate to the task. However, I was more interested measuring how fast the B12S could perform and how much energy it would use.

For that series of tests, I used SANWatch to create logical drives, each containing six physical drives in RAID5 with no spare. In a typical setting, a spare disk would, of course, be a good idea, but I wanted to make sure that all drives were spinning and using power during the test.

I assigned each logical drive to one of my servers, then I began running Iometer scripts to stress the number of I/Os per second and the transfer rate. During those test runs, my Watts Up Pro hardware meter was recording the power used by the array only.

Performance-wise, the EonStor B12S fared well, rivaling results I've seen from arrays with 3.5-inch drives that use similar technology. This didn't surprise me in the least. I was, however, surprised by just how low the B12S's power consumption proved to be. When idle (i.e., while neither reading nor writing data), the B12S consumed a mere 235 watts, remarkably less than the 380 watts I have measured on arrays with 3.5-inch drives and similarly redundant, six-drive configurations. Another interesting fact is that the workload doesn't add much to the baseline, only a few watts more. In fact, the increment is almost perfectly linear with the block size increase, with a few spikes when the cooling modules of the array start spinning the fans faster, which happened, for example, when the room temperature would temporarily drop a degree or two below 75 F. Even at the highest stress level I measured -- during sequential writes of 64K blocks -- the B12S drew a mere 254 watts. See my Iometer results at various block sizes and their corresponding power consumption figures.

Infortrend offers an interesting approach to pricing: Unlike other vendors' arrays, Infortrend's enclosure doesn't mount drives with proprietary connections. Thus, you can purchase the enclosure (the one I tested costs just under US$8,000), then procure your own drives separately.

I enjoyed reviewing the Infortrend EonStor B12S and I would not hesitate to recommend it for consideration for shops where its capacity and other features fit your requirements. Boasting performance in the same ballpark as traditional arrays, a power draw that is roughly 60 per cent less, and an overall form factor just half the size, the B12S can be an offer too good to pass on.