Fake 4G networks and crippled column databases
There’s a recent trend among North American mobile carriers to re-brand their 3G cellular networks as “4G”. Originally meant to classify networks approaching 100 Mbps, Sprint began using “4G” for their WiMax network and now – comically – AT&T for their barely 14.4 Mbps HSPA+ phones. Why? It’s easier to print “4G” on the box than to build an actual 4G network. It’s the same like of thinking that leads to painting racing stripes on the side of a Camry and expecting it to “Go Faster.”
Anyone else notice the parallels in the analytic database industry?
Some traditional transactional vendors, Oracle for example, have taken the AT&T approach and tried to bolt on column databases hoping their customers confuse the name with the performance. Others have tried, like Sprint, to deploy fringe technologies that offer some benefits over old approaches but little or no future beyond them. And a few have taken the first steps towards column store not full implementations, but compared to the horrible performance of everyone else it can seem fast… until you compare it with something best of breed.
SAND has spent decades perfecting a column oriented, tokenized, bit array-encoded database engine that truly offers high performance for tens of thousands of concurrent users on hundreds of terabytes of data doing the toughest complex queries imaginable. Why? It’s better to build an actual next generation analytic engine than simply add a bullet point to a website. We didn’t paint any stripes on the car or stick any extra Gs on the box — we really made it to go faster.
Sure, it’s much easier to just talk about “column store” than to spend a couple decades proving results with actual customers. In apples-to-apples comparisons in real world settings, true LTE 4G networks will blow away the currently marketed pseudo-4G phones we have today. In apples-to-apples comparisons, with real data, on-site, and in ask-anything tests, true analytic databases like SAND will simply blow you away.