Exasol: Taking a bet on the affordability of in-memory analytics

Exasol has beaten a separate path from rivals in the market, and while open-source systems have climbed in popularity, the German database biz has remained proprietary. And although other vendors have tried to bring together analytics and transaction in one database, Exasol remains, steadfastly, an analytical system. This week …

[spireite]

Name....

For years, I always wondered why someone would name their company like a suppository.....

[Korev]

Re: Name....

Where can I send the invoice? -->

[Korev]

Thinking of machines with ~1000 CPUs in the 90s is incredibly far-thinking.

The only single machine I can think of with that kind of scale were the Cray XMTs. I think CSCS' machine had 2TB of RAM and 8192 threads. I guess we'll see "normal" servers with that scale towards the end of the decade.

A pint of Britain's finest for the German boffins -->

[Steve Channell]

Teradata could scale at a price

Teradata was/is a massively scalable solution using hash distribution of data connected by a proprietary network interconnect: the problem was cost, and nobody had a problem big enough problem to justify the spend to scale to thousands of nodes

[Anonymous Coward]

"...were still 32-bit so you could only address 2GB of main memory."

You must be referring to a DIMM or a CPU/FSB design limitation or forced signed values within the OS or something other than 32-bit addressing to 2^32 int values (4294967296 bytes/ints). That, or either you meant 16-bit or 4GB.

For the article, I bet they never imagined that memory capacity would stall out at such low values, not when CPU's in that age were doubling in speed every other month (seemed like). You can't find today a 1200MHZ system CPU but, you can find a system with a 8GB RAM, and both were considered a lot 20 years ago.

[Steve Channell]

2GB was a common limit at the time because the OS reserved an area of memory for the kernel and memory mapped devices.