Whites Papers

In the mid-90s R+ was literally measuring nothing, which was ultimately what it was always intended to do. To understand what R+ accomplished, go back in time to the late 60s, early 70s when abnormal end of production jobs, or abends, were the scourge of the IT industry. In the 60s, most IT organizations did not measure the failure rates of their hardware, so they had no way of knowing if those failures were increasing or decreasing. Had they been measuring those rates and found that those failures were increasing, it is not clear what they could have done about it. Then we introduced R+ in 1970, and IT management learned firsthand the power of an industry-wide management measurement system…the industry’s first “collective measurement" system. The biggest change that R+ pioneered was the idea of getting first dozens, and then hundreds, and finally thousands of data centers to measure their abends using the same measurement protocol. Then we collected and pooled the results of all those measurements into an industry data base against which the measurement of any one data center and any one vendor’s hardware could be compared. This is what we termed “collective measuring”.

Within a year of the introduction of R+, Telex, the largest of the plug compatible vendors was on its way out of the industry because the R+ collective measurements showed the reliability of its products were substandard. Within a few more years CDC and Memorex also departed the industry for similar reasons. But the loss of these three vendors did not lessen competition. The surviving vendors started making major improvements in the reliability of their products in response to the R+ "collective measurements" and rankings. Over time, vendors began to build hardware specifically designed to prevent abends. The team that pioneered RAID technology acknowledged that one of their major reasons for their design was to do well in the R+ rankings by making sure that their hardware never caused abends. 

R+ demonstrated to hardware users, that being big did not necessarily mean being good. Some of the largest IT organizations in the US found themselves ranked by R+ in the bottom half of the industry. One of the largest financial organizations in the US found that despite a policy of only using IBM hardware, having its CEO on IBM’s board and having IBM’s CEO on its board, its hardware reliability, when ranked by R+, was among the worst in the industry. Two of the largest IT organizations in California learned in the same week that their size did not protect their hardware from also being ranked in the bottom 10% of the industry.

The original R+ demonstrated that combining internal measurements collected from thousands of data centers using the same measuring protocol could achieve extraordinary results that internal measurements used only internally would never have accomplished. In the mainframe world, abends went from millions a year to virtually none. Now RPI proposes to focus that "collective measurement" concept on the events that disrupt the network system with the goal of reducing those events from hundreds of millions a year to none. Internal IT measurements can tell IT management where it is and where it has been, but only RPI’s "collective measurements" can tell IT management where it could be and explain how to get there. Only collective measuring can trigger the kind of competition between vendors of goods and services that will lead this industry to an interruption-free network system.