The thing I keep finding confusing about these companies is: did anyone who cares about performance bother coding their app to use MySQL in the first place (as opposed to a database like PostgreSQL or Oracle, which have better index search and lock algorithms)?<p>This is really important, as MySQL has a weird dialect of SQL with all kinds of broken MySQL-isms (including fundamental things like not using the SQL-standard meaning for the various forms of quoting) as well as drastically limited support for supposedly-advanced things like sub-selects and truly-advanced things like windowing and partition functions...<p>Seriously: I don't think I have any apps that would actually /run/ on something that spoke MySQL without recoding all of the data access parts (and that's pretty much what a web backend is) from scratch, swearing and cursing the entire time about why I don't get to use X or Y or Z feature.<p>To me, this is just "yet another of those MySQL optimizer companies" that we see come and go year in and year out, and one that is almost certainly going to experience the same fate as all of the other ones: they are a dime a dozen, and almost always fail. The failure rate on PostgreSQL optimizer companies, however, seems much lower, despite the fact that they don't seem to get the fancy funding (probably because, if you don't take into account the underlying demographic of "performance-savy" from "database user" it looks like a smaller market).
Not a lot of info on their website. But it strikes me as similar to VoltDB (<a href="http://voltdb.com/" rel="nofollow">http://voltdb.com/</a>). Especially the emphasis on OLTP queries.<p>Is it based on the same paradigm of single-threaded serial access? What differentiates it from VoltDB? Is it the MySQL compatibility and ability to use ad-hoc queries instead of stored procedures? (If so, how do you make that bit fast?)
Hey, guys. Congrats on the funding.<p>Why would I use MemSQL as opposed to relational database on SSD? RDMS on SSD gets me a similar speed boost as opposed to running on a completely new database.
How MemSQL compares with TimesTen? Or even with Oracle RDBMS with bunch of RAM? Or Oracle RDBMS with Oracle In-Memory Database Cache (TimesTen) on application app servers?
What is MemSQL "secret sause" comparing to competitor?<p>EDIT: Just so say that I'm so very happy to see a company which tries to solve real problems not Web 2.0 social networking crap is able to raise money.<p>EDIT2: The company like this needs $40M investment not $2M.
A real victory for YC as well as the founders (congrats to them - great software!)<p>This is the best example yet of YC's considerable ability to take a very deep tech, non-consumer startup, help vet and guide its technology and marketability, and provide the credibility to get it well-funded.<p>Fantastic round for a database software startup, a breed written off not long ago. I'd think it wouldn't happen maybe anywhere besides YC. Really shows why YC should be (and is) sought out by founders looking to build this kind of company.
I never heard of a database that doesn't reside in memory if you have enough ram. Most of the time, it's the OS making this happen by maintaining a cache for disk reads, and flushing writes to disk asynchronously.<p>Does MemSQL just have more compact data structures that let more stuff fit in memory?
When I hear about such "MySQL replacement" databases built from grounds up by companies that are just a few months old, I always wonder how complete their support for MySQL feature set is. Do they support every SQL feature/quirk/misfeature of MySQL?<p>It seems to me that unless it is a complete one-to-one mapping, it is going to be a nightmare to explain what is supported and what is not.
How does performance compare to MySQL running from a RAM disk?<p>(I can imagine designing for RAM from day one can help... but curious as to how much it helps, once the actual storage layer and investment in RAM is identical.)
Ok - so i posted 2 questions:<p>1) Can @memsql scale writes ?<p>2) Is @memsql ACID compliant ?<p>Your answering on sharding indiciates that @memsql scales writes in exactly the same way as MYSQL - sharding<p>Re. ACID complaince. I'm guessing you are ?