In our latest IT Pro Ranking, we asked users of midrange storage arrays to tell us which products they like best and why. We received enough survey responses on six vendors to ensure statistical validity, and Dell/Compellent was the clear favorite.

The other five in our ranking are IBM, NetApp, EMC, Dell/EqualLogic, and Hewlett-Packard. (We offered a list of 15 vendors but received enough responses for only six to be considered in our final report.)

Dell/Compellent and IBM tied for No. 1 in our ranking's standardized evaluation criteria (more on those later), while Dell/Compellent and NetApp tied for the top spot in our storage array features evaluation.

EMC ranks highly on virtualization integration and performance, but not so well on port density and data deduplication. Dell/Compellent scores high on reliability and performance, but not so well on breadth of product line and service innovation.

Dell's EqualLogic line fares well on acquisition costs but lags in innovation. IBM scores big on product reliability, but it takes its knocks on operation and acquisition costs. NetApp does fine on reliability, but it's right down there with IBM on operation and acquisition costs. HP comes in as an also-ran: It does well in reliability and performance, where it should, just not as well as the leaders.

With data protection, high availability, and virtualization high on survey respondents' priority list, you'd expect that EMC would have finished higher in our ranking of product-specific features. But users rate it near the bottom of the pack for data protection. NetApp and Dell/Compellent rank the highest for data protection.

One of the most important product features in determining vendor rankings is deduplication. That's not because respondents rank that feature as very important--it ranks in the middle--but because most vendors don't rank well on built-in dedupe. HP, EMC, and Dell/EqualLogic score below 3.6 on our 5-point scale for dedupe.

Hands-On Evaluations

A quick reminder of how the InformationWeek IT Pro Rankings work. Only IT pros who have used or done hands-on evaluations of the products in our survey are allowed to grade them, so on the whole these assessments tend to trail the feature sets of the latest offerings. So if a vendor greatly improved or added a feature, or substantially dropped a price, that's something Gartner analysts could easily fix in their assessments, but we can't.

Users are asked to evaluate the products on both a standardized scale (which includes product performance, reliability, purchase cost, operation cost, and breadth of product line) and a product-specific scale (data protection, high availability architecture, virtualization support, and 10 others). But they're also asked to rank the evaluation criteria themselves. By using this methodology, even if we include a dumb evaluation criterion, users will just rank it as inconsequential.

Our goal is to keep "experts" and vendors out of the mix and let your peers speak, in a statistically defensible way, about the products they use. Vendors in the rankings don't know that the survey is happening and have no ability to opt in or out of it.

We've been doing these surveys for more than a year now. One outcome we hadn't thought of is that by letting participants rank both the criteria and products against those criteria, we end up with vendors ranked very close together. In most cases, product users are usually happy with their product choices, so when an IBM product ranks high for reliability and low for acquisition cost, that finding tends to be reflected in the evaluation criteria. In this survey, users rank reliability as their No. 1 criterion and acquisition cost in the middle. Thus, only three percentage points separate the field on one scale, and five points on the other.

The bottom line in our latest IT Pro Ranking is that you're generally getting what you expected from your storage vendors, so very small variations in our data indicate where vendors are either delighting or disappointing customers.

Check out what your peers think of their midrange storage arrays at InformationWeek Reports.