SATA Flash Solid State Disk up to 160 Gbyte announced!

Mon, 2007-02-26 09:33 — Shinguz

The price for a 160 Gbyte disk will be around USD 15'000. This is still a bit expensive. But the access time is around 0.5 ms (both for reading and writing) which is around 10 times faster than a normal 15'000 rpm SCSI disk! The disk has NO cache because it is a cache itself (according to the supplier. Maybe this will change in the future). And the lifetime of a cell is > 5 mio writes. For the same performance one needs usually an array of around 10 disks. If your database is heavily write-I/O bound you should consider this solution.

I/O-Systems

Will NAND Flash / Solid State Disk (SSD) will be the future for Database I/O systems?

Actually SSD are still limited in size (64 GB) and expensive (EUR 30/GB) and thus cannot yet compete with SCSI or IDE disks. But they also have an advantage. They are fast!

For some uses like databases the price per GB is not that relevant. Also most of the databases fit into one or two 64 GB disks (more than 90%).

When you really get an I/O bottleneck you should consider to choose SSD:

The demo box uses a 32GB 1.8-inch solid-state drive created by Samsung, to show off the company's flash chops. Unfortunately, at $30 per gigabyte, the 32GB drive would cost over $900.

When a user requests that data, rather than being limited to servicing 100-200 requests per second (as with a traditional HDD), Samsung's SSD can service up to 5000 request per second, virtually eliminating data seek delays.

Chip-Test:

Read avg.	53.5 MB/s
Write avg.	26.4 MB/s
Access read	0.2 ms
Access write	36.0 ms

A Comparison of Solid State Drives to Serial ATA (SATA) Drive Arrays for Use with Oracle gives: 30% faster in write and 10 times faster in read!

Adtron announces SATA Flash Solid State Disk up to 160 Gbyte!

2007-02-26: Adtron A25FB-20 Flashpak®

The disk has NO cache because it is a cache itself (according to the supplier. Maybe this will change in the future). And the lifetime of a cell is > 5 mio writes.

For the same performance one needs usually an array of around 10 disks.

If your database is heavily write-I/O bound you should consider this solution.