This web site is nowadays much slower than a few months ago. Can anything be done to improve performance. Usually I go to another room to do somthing else after each click hoping it has responded when I come back.
thanks,
Panu
Yes, I agree. C2 has really slowed down over the past month or two, as the number of users has increased.
To fix the problem, we’ll be adding some new servers (and slightly different database architecture). This should improve performance. Hang in there until the upgrade is done. I’ll keep everyone posted on our progress.
Matthew
You may want to order this white paper (it’s free):
http://sleepycat2.inetu.net/offers/newcombe/index.php?From=bdbpp
… and take a look at the documents on this page:
http://www.oracle.com/database/berkeley-db/db/index.html
If you don’t need structured SQL calls, but do need fast indexed data retrieval, a database that requires little administration, ACID compliance, atomic operations, live replication and mirroring, and hot-standby master failover, this particular [free] product will excel at that.
Don’t know if my comments are considered helpful or meddling… assuming the former, here’s another hint:
I have found multicast to be of great help in the design of my automation platform. The primary feeds server… responsible for maintaining a connection to, receiving and standardizing feed data from various sources… makes this data available to the entire cluster via one-hop multicast (which prevents it from being broadcast outside of my network). Any of the other machines in the cluster (signal server, trade execution server, trade management server, newsfeed server) can then receive this “standardized” or “normalized” aggregated feed by joining the multicast group. By doing it this way, I only need to broadcast the data once (at which point it is in a standardized format regardless of provider) regardless of how many machines need the data (which means less bandwidth overhead).
Your suggestions aren’t meddling, and this approach you suggest is very interesting.
But I’m not sure that the bandwidth between the feed(s) and the other C2 servers is really the source of the problem currently. The primary problem today is the database architecture. So I need to solve that first before optimizing other stuff.
Yes, I understand your problem is centered around data access and retrieval. My first comment regarding SleepyCat/Oracle Berkeley DB applies, if it fits into your programming environment and if you can do without structured SQL calls (which many applications really don’t need).
I’m mentioning architecture-related comments just as an F.Y.I., because I generally design in a manner that avoids future problems. I do not want to have to solve an architecture-related problem with my system five years from now, because I don’t think my mental faculties will be fully functional by then. I also want to design a self-healing system that will continue trading for my wife should age take its toll on me.
It is ten times the work to design a system that anticipates certain problems and executes recovery (self-healing) routines, but the nice thing about it is you only have to do it once - and failures generally don’t interrupt your life (or vacation) ever again.
It’s ironic, in a way, that this is probably the last “big” project that I’ll be working on… and that it ends up being a personal one, at that. By “big” I’m referring to system, architecture and redundancy complexity. It is fun, though… and very satisfying to see its results (in terms of profitable trades).
