Groovy Corp puts out a ridiculous press release
I knew Groovy Corp’s press release today would be bad, as it was pitched in advance as being about an awe-inspiring benchmark. That part met my very low expectations, emphasizing how the Groovy SQL Switch massively outperformed MySQL* in a benchmark, and how this supposedly shows the Groovy SQL Switch would outperform every other competitive RDBMS by at least similar margins.
*While a few use cases are exceptions, being “better than MySQL” for a DBMS is basically like being “better than Pabst Blue Ribbon” for a beer. Unless price is your top consideration, why are you even making the comparison?
Even worse, the press release, from its subhead and very first sentence, emphasizes the claim “the Groovy SQL Switch’s ability to significantly outperform relational databases.” As CEO Joe Ward quickly agreed by email, that’s not accurate. As you would expect from the “SQL” in its name, the Groovy SQL Switch is just as relational as the products it’s being contrasted to. Unfortunately for Joe, who I gather aspires to edit it to say something more sensible, the press release is out already in multiple places.
More favorably, Renee Blodgett has a short, laudatory post about Groovy, with some kind of embedded video.
Comments
17 Responses to “Groovy Corp puts out a ridiculous press release”
Leave a Reply
[…] following was originally part of my post today regarding Groovy Corp, but I decided to post it separately […]
My takeaway from that press release is that Groovy claims that MySQL is the fastest relational DBMS. Or perhaps they mean it was given this new result on the OLTP benchmark for which all vendors publish results.
Presumably right, Mark. They think MySQL is the standard to beat. Of course, when your product runs best on a 24-core SMP box, that’s not a very defensible stance to take.
Renee Blodgett is a PR spokesperson. Probably shouldn’t take her word for it.
http://www.spock.com/renee-blodgett
I don’t even take the word of professional analysts I admire.
But I’d be surprised if Renee were getting business from Groovy Corp right now.
And while I was alarmed to see that she’d recently written about “pay to post” arrangements with bloggers, I was relieved to see that her take on them was negative.
This is an awful press release, but not just because it talks about MySQL as the fastest.
You can make MySQL pretty fast by using a storage model without transactions. This kind of model is often acceptable for really high performance applications. A couple of years ago I ran a test of such a configuration against the few RDMBS vendors that support a transaction-free mode and got pretty good performance from MySQL, maybe marginally better.
So – to kick a standard Oracle configuration’s butt, just turn off transactions in your RDBMS. And if possible, use the OS file cache and compare it to an RDBMS that doesn’t. You get streaming inserts that never, ever interfere with queries, which is about 90% of the problem with most RDBMS under super-high-performance.
What would it take to beat MySQL? How about a bigger memory cache? Or a more efficient memory cache, that can do its own queries?
An interesting feature, but hardly revolutionaty, and definitely not right for every application.
Bottom line – unless you know exactly what features you get along with your high performance, you don’t really know if the sacrifices are worth it.
Hans,
It’s a RAM-based DBMS proving the “surprising” fact that it beats a disk-based DBMS’s performance all to hell.
What about apple/coconut comparisons don’t you like? 😉
I don’t get it. Groovy is focusing on databases for etailers. It picks a seemingly well respected benchmark from Dell that compared relational databases, and shows it is x times faster than the fastest of those.
The only reason MySQL comes into the picture is that it was the previous best one. I’m sure they would have preferred to compare with Oracle, but that’s not what the Dell benchmark showed.
If other RAM-based vendors can show better performance than them, it’s an open-source benchmark and they’re welcome to do so. Or if Curt thinks the Dell benchmark is nonsense, he should explain why or suggest an alternative.
But I don’t understand the reason for all the heat.
I don’t like the misleading and otherwise confusing language of the press release.
I don’t actually have an opinion on the Dell benchmark per se.
I think benchmark comparisons of RAM-based vs. disk-based DBMS are really stupid.
Your comment about Groovy’s supposed market focus doesn’t happen to jive with what CEO Joe Ward told me a couple of days ago, but I agree that would seem to be an obvious fit for their technology.
Hmm, ok I read over their site and I actually sort of liked some of it. But They look a little unfocused. Check out the worst product information page, ever. Probably trying to find the niche for their mix of technologies. They should definitely be careful about how they go about getting attention.
They may have something real — but boy are they doing a bad job of building a marketing base for it. The sad part is that because they’re covered on TechCrunch and here and so on, they may not realize they’re doing badly.
Not sure if that’s the worst product page EVER, but having a product pitch mixed in gets major — well, I’d say red marks, but on an almost unreadably bright orange page, who would notice those?
If I read the press release correctly, Groovy is an in-memory database, sort of like SolidDB. Of course it’s faster! They’re comparing apples and oranges, as it sounds like the MySQL setup is a “standard” disk-based DB.
David,
You’ve nailed it.
Appreciate your followup to my comment, Curt! Can you disclose who they are targeting? I must have confused them with somebody else when thinking etailers.
I’m not sure why you say RAM vs disk-based benchmarks are really stupid? Disk is what everybody has right now, so it seems like a natural point of comparison to get customers thinking about the merits of their current technology platform. I do understand that it becomes easy to make very significant claims, but that’s partly the point of showing the basic advantage of technology approach, no? I would think the key thing right now is to show why they’re orders of magnitude better than the incumbent technology, not some fraction better than some other startup approach. What’s your take on this?
I actually appreciated their use of an open-source benchmark, since some one else can show why Groovy’s not the best RAM-based option with the same benchmark.
I hate benchmark announcements that focus on detailed numbers with little explanation of why the results were achieved, at least beyond regurgitations of generalized marketing hype. And Groovy doesn’t even get the generalized part right. If they’d said “RAM clobbers disk, as our benchmarks show, and we have the best/only viable RAM-based solution because of X, Y, and Z”, that would have been much better.
(Sorry for the late post.) I just read several paragraphs of their web site and I do not understand what this is about. Like most people, I don’t have time to keep going without even knowing whether this is something I have a need to know. Curt, I hope you’ve expressed to these people how bad they are doing at explaining themselves. For all I know, they might have something wonderful here, but nobody would ever know it. I should think they’d listen to your advice and take it seriously. And if they don’t, that’s a pretty good indication that they aren’t savvy businesspeople.
[…] Groovy Corporation seems to have disappeared, or morphed into something called uCirrus, or something like that. Categories: Akiban, Cassandra, Clustering, Clustrix, Code Futures and dbShards, Facebook, Groovy Corporation, NoSQL, Northscale, OLTP, Parallelization, ScaleDB, Specific users, VoltDB and H-Store Subscribe to our complete feed! […]