Recently, I've had the opportunity to install a solid state storage device for a customer. (Think of your USB thumb drive, or the card you shove into your digital camera, only this thing plugs into your server's 4x PCI slot.)
At first glance, the thing is fast, really really really fast. How fast? How about +500MB/s write speeds?
That's turning it way past 11, folks...
Historically, many of the servers and arrays our customers have installed have been a decent investment in hardware - $5-20K+ easily. These usually consist of a beefy server, racks, installations, multiple disk arrays, array controllers, and considerable OS tuning and tweaking to get 100-200MB/s out of the investment. Add to that some time to install postgres, config it, tune it, load it, test it, etc.
What does this mean for the larger xTuple installations out there?
How about lower power consumption in your datacenter? Faster backup/restore times, perchance? What about spending less money on servers? Maybe you'd like to put an end to most dreaded question - 'What is taking so long?'
Below are some times for a standard write speed test that I run:
Here's a result from an Xserve running OSX - 3 disks in RAID 5 (RAID 5 not recommended, BTW)
OSXserver: root# time bash -c "dd if=/dev/zero of=bigtestfile bs=8k count=512000 && sync"
512000+0 records in 512000+0 records out 4194304000 bytes transferred in 72.526625 secs (57831231 bytes/sec) real 1m13.037s
Here's a result from a HP server.
root@HPserver:# time bash -c "dd if=/dev/zero of=bigtestfile bs=8k count=512000 && sync"
512000+0 records in
512000+0 records out
4194304000 bytes (4.2 GB) copied, 68.7805 s, 61.0 MB/s real 1m53.352s
Here is the SSD card result - _37GB_ in _69_ seconds...
[root@xTuple fio2]# time bash -c "dd if=/dev/zero of=bigtestfile bs=8k count=4469740 && sync"
4469740+0 records in
4469740+0 records out
36616110080 bytes (37 GB) copied, 68.8079 seconds, 532 MB/s real 1m9.471s
This is seriously Playing-Poker-With-The-Rat-Pack cool...