Bitcoinica System Status

[zhoutong]

We're really proud of our robust technologies that make your trading experience wonderful. Here we showcase how much traffic our servers are handling right now and how responsive our services are.

https://www.bitcoinica.com/pages/status

Currently we have handling almost 10,000 requests per minute and average response time is around 10ms. Pretty impressive, huh?

[1713482357]

1713482357

[1713482357]

1713482357

[1713482357]

1713482357

[1713482357]

1713482357

[1713482357]

1713482357

[tvbcof]

We're really proud of our robust technologies that make your trading experience wonderful. Here we showcase how much traffic our servers are handling right now and how responsive our services are.

https://www.bitcoinica.com/pages/status

Currently we have handling almost 10,000 requests per minute and average response time is around 10ms. Pretty impressive, huh?

You really have not given enough info to judge how impressive things are to be frank.  200qps is actually not gigantically impressive in a lot of situations, and 10ms measured at the front of your app seems ho-hum to me, particularly if you are making good use of redis.

The challenge is going to be providing service close to the edge (how well you do replication if you choose to attempt it), and how seemlessly you can handle failover events and whatever other speed bumps the real world will drop in front of you.

Lest you get me wrong, I am impressed with what I can see of your work (which exceeds my own skill level by orders of magnitude) and with what I understand of how it was accomplished, but I suspect that a lot of challenges lay in front of you.

[phorensic]

Uh, yes, very impressive.  I was limited to a single spindle of 7200rpm HDD on our Mining pool and I couldn't get my SQL db response below 10ms even with everything supposedly cached in RAM.  I realize that was a really crappy hardware setup for a SQL server, but are you using SSD?

Edit:  I work with and have tuned the shit out of a 24 spindle NetApp system at work, so I'm not new to DB and disk performance measuring.

[zhoutong]

Uh, yes, very impressive.  I was limited to a single spindle of 7200rpm HDD on our Mining pool and I couldn't get my SQL db response below 10ms even with everything supposedly cached in RAM.  I realize that was a really crappy hardware setup for a SQL server, but are you using SSD?

Edit:  I work with and have tuned the shit out of a 24 spindle NetApp system at work, so I'm not new to DB and disk performance measuring.

I'm using Rackspace Cloud. The private networks, I/O and CPU are the awesome parts. Actually I'm handling this amount of traffic with 2 web servers utilizing 500 MB RAM each, and MySQL database is only utilizing 120 MB RAM. (Redis is negligible.)

So I think it really has to do with the actual code quality and optimization. Hardware is important too, but throwing money at servers is always easy.

I'm a n00b in scaling by the way.