I have several million 15-minute electric usage records for which I need to determine an appropriate schema and partition columns for VoltDB. I'm trying to determine some throughput benchmarks.
With some other databases, I've used a tall table, where I would represent the readings for a single day with 96 rows. This ended-up causing some pain with a few hundred million rows. So, I went to a wide table, where each day of 15-min readings was represented by 1 row, with 96+ columns. This eased some pain for a while, but I've had some bottlenecks again.
Hence, trying VoltDB.
* For a high performance database such as VoltDB, does either design wide vs. tall, lend itself better to VoltDB?
* Does providing 96+ input parameters to a volt procedure pose a problem?
I'm sure I'll have additional thoughts on planning for capacity, memory, number of servers, etc. But, I have to start somewhere...
Thanks
With some other databases, I've used a tall table, where I would represent the readings for a single day with 96 rows. This ended-up causing some pain with a few hundred million rows. So, I went to a wide table, where each day of 15-min readings was represented by 1 row, with 96+ columns. This eased some pain for a while, but I've had some bottlenecks again.
Hence, trying VoltDB.
* For a high performance database such as VoltDB, does either design wide vs. tall, lend itself better to VoltDB?
* Does providing 96+ input parameters to a volt procedure pose a problem?
I'm sure I'll have additional thoughts on planning for capacity, memory, number of servers, etc. But, I have to start somewhere...
Thanks
Comment