Eliminating the DeWitt Clause for Greater Transparency in Benchmarking


Jack Chen

Software Engineering Manager

Eliminating the DeWitt Clause for Greater Transparency in Benchmarking

SingleStore is eliminating the DeWitt Clause from our contract, allowing us to enable greater transparency around database performance among ourselves and industry competitors.

SingleStore always strives to push the boundaries of database performance, and we love sharing the technology  we've built to accomplish that — not to mention, helping our customers and others in the database industry understand the technical innovations that enable SingleStore to achieve state-of-the-art performance on both OLTP and OLAP workloads, in the cloud and at scale.

But we don't want you to take our word for it. We believe anyone should be able to test and benchmark SingleStore and see the results for themselves, and anyone should be able to openly discuss and compare different database products — not only help customers make the best decisions, but also help the industry recognize the innovators who push the boundaries of what databases can accomplish.

Sadly, the industry has a history of closing these discussions off by forbidding users from publishing benchmarks on their database products. This practice originated in the 1980s, when Professor David DeWitt wrote a database benchmark and tested it on several databases (including Oracle), uncovering weaknesses and poor performance. Oracle responded by adding a clause to their licensing agreement that prevented anyone from publishing benchmarks without their approval — effectively disallowing anyone who finds negative results to openly publish them.

But it wasn't just Oracle. Nearly every database vendor followed suit — and this clause, known as the DeWitt clause, became the de facto standard throughout the database industry, stifling open competition and innovation. It's why so many benchmarks (including our own) often compare against anonymous products called "DB-A," or lack comparisons entirely.

When SingleStore started 10 years ago, we also included a DeWitt clause, following industry standards at the time.  But we're glad to see the industry standard shift in recent years, with more companies and groups like IBM, Azure, AWS, Databricks and Snowflake supporting open and transparent benchmarking.  The DeWitt clause doesn't match our philosophy, we've never made use of the clause and we never want to.

As a result, we’re updating our terms to eliminate the DeWitt Clause from SingleStore. And like many of the database vendors listed above, we’re implementing a "DeWitt Openness Clause" to require that if a competitor benchmarks SingleStore, then they must also open their database for us to benchmark them, negating the ability of a competitor's DeWitt Clause to stifle competition solely for their benefit. And these benchmarks must be transparent and reproducible, so that any reader can review the results themselves.

We believe our users should be able to benchmark SingleStore and openly discuss the results, good or bad. We believe customers should have access to as much information as possible to evaluate database products on their merits. And, we believe benchmarks should be transparent so others can reproduce the results and adapt the methods to their needs. This transparency allows the database industry to better understand its weaknesses and develop innovations to turn them into strengths, rather than hiding from those weaknesses. And we're glad to see that we're not alone in these beliefs — the industry is starting to move in a new, more open and transparent direction.

We let our performance speak for itself. Try SingleStore for free today.