In-memory database vendors Hazelcast and GridGain trade benchmarking blows
Greg Luck alleged that GridGain – which is a commercial implementation of the Apache Ignite open-source project – configured the tests differently when it published a direct comparison of the two products. That comparison was itself a response to earlier tests published by Hazelcast.
+ ALSO ON NETWORK WORLD:Business leaders still in denial about cybersecurity threats | Opera founder talks up new Vivaldi browser+
“We believe this behavior is unbecoming of an Apache project,” Luck wrote, adding that GridGain may have violated the Apache Incubator code of conduct. (Ignite is a recent graduate of the incubator.)
GridGain co-founder and executive vice president of engineering Dmitriy Setrakyan said that his company had performed the benchmarks again, and that the results were not only precisely what they were the first time around, but that they were independently verifiable.
“Our benchmarks were run on Amazon AWS, the code is published, and anyone can retry them,” he told Network World via email, noting that Hazelcast had altered the testing parameters and run the benchmarks on in-house hardware for its rebuttal.
Setrakyan added that GridGain had reached out via user group postings to Hazelcast, sharing their results and asking for a review – without any success.
“Greg Luck has removed our post and provided no further commentary,” he said.
Both Hazelcast and GridGain are at least partially open-source, and the two companies apparently regard each other as bitter rivals in the in-memory database space. IMDBs are what they sound like – database frameworks that use main memory, instead of traditional storage, to hold information. They’re consequently a lot faster than regular databases, and useful for any application that has to have near-real-time responsiveness.
GridGain, on the same day as Luck’s accusatory blog post was published, announced that it had closed a $15 million series B funding round.
Both companies, it appears, have more to say on the matter. Via a spokesperson, Luck said that Hazelcast is looking into further issues with GridGain’s benchmarking. “Our performance team is digging deeper across the entire benchmark suite to see if there are other problems,” he said.
For his part, Setrakyan promised further details in a forthcoming blog post.