Scalability issues

In a thread posted on July 27, 2012, grarpamp reported that a certain class of machine had become useless for Bitcoin. The issue arose when blk0002.dat was created to store more blocks, causing forward progress processing blocks to turn into losing ground by approximately 20 a day due to disk-based overload resulting from both datfiles being accessed at once. There were no mentions of crypto in the thread except for this post, so it is unknown how new hardware would perform under these conditions. Another user in the thread mentioned having an 1.8 GHz P3 Celeron (128k cache) running vintage 2.6.20 Linux, which should be substantially slower than grarpamp's machine; however, they reported staying current with syncing the chain just fine. They did not log the time it took to sync as they forgot to turn on timestamp logging, but it was less than two days as that was the span between when they checked on it. The user encouraged grarpamp to investigate their software configuration.