Welcome to NexusFi: the best trading community on the planet, with over 150,000 members Sign Up Now for Free
Genuine reviews from real traders, not fake reviews from stealth vendors
Quality education from leading professional traders
We are a friendly, helpful, and positive community
We do not tolerate rude behavior, trolling, or vendors advertising in posts
We are here to help, just let us know what you need
You'll need to register in order to view the content of the threads and start contributing to our community. It's free for basic access, or support us by becoming an Elite Member -- see if you qualify for a discount below.
-- Big Mike, Site Administrator
(If you already have an account, login at the top of the page)
I hear you bro! There is a NT strategy I'm starting code that phones home... literally. If a remote NT session loses connectivity to me, it gracefully flattens everything.
Got that tip from a friend in the HF space who swears its better not to fly at all then to fly blind.
<a bit off-topic>
This is close to what I used for high-availibility clusters: server #1 send its status to server #2, and #2 to #1. If a failover is needed, and server #2 don't know in which state is server #1, it will power-off or reset it, it's called STONITH (Shoot The Other Node In The Head), to avoid dual-brain syndrone (each machine thinks the other one is dead, so starting the same applications, mounting the same disks, etc, very bad idea...).
In the real world, after few hundreds of failover, the conclusion is that was a good idea.
So it's not the same application, but I agree 100%, "better not fly at all than flying blind" .
All very good suggestions ... I have that type of setup + actually have 4 UPS units ganged together + test them monthly as well
... except I need to go to the step of having another machine loaded with NT and mirror the working machine ... I have already (months ago mind you) had to call my broker twice due to a localized problem, to flatten my account ... now I dont monitor ES anymore and I dont have NT freeze up on me when the markets go crazy .... oh welll ... nothing lost
Could give more information about your setup. If I understand, your application is running using MC, Ninja or another software on an outside, virtual server.
Very interesting as I'm afraid to autotrade from home. I'm not gear like those proffessional servers
I have half a rack in a data center which is pretty empty since the last server refurbishing for now.
FOR NOW my data center side server consists of:
* An old server (4gb, dual core, 4 discs) that serves are physical domain controller and management point
* A low midrange server consisting of 2x4 core opterons, 64gb RAM with a currently 12 disc subsystem running Hyper-V (january planned upgrade: another 5 discs).
* About 2 dozen virtual machines, among them some windows 7 desktops (development, trading), some hosting insfrastructure, exchagne etc. but also some machines that form the backbone of a market database.
Plus a switch, a layer 3 switch (hardware based router) and empty space.
THe main server is slowly filling up the memory, so I will soonish get another server (1 rack unit, 2 nodes, each 2x4 cores, 64gb ram).
In the office I have another 2 servers running on AMD desktop chips with soon both 16gb RAM and 8 discs each forming my development backbone (code archve, build servers etc.) as well as quite some data archive.
Note that my "trading operation" is attached to a small IT company and the infrastructure is merged. SO, some of the stuff actually are our internal email and video conferencing system etc.
Basically,
* Get a server
* Put in virtualization
* Install computers into this.
@NetTecture whats your view about having a VM facing your broker/clearing engine vs. dedicated hardware for medium frequency stuff (75 R/Ts a day on CME)?
It depends on your strategies. VM's react slower due to the permanent proces switching. We talk of delays in the range of 55ms here.
I fyou try to do HFT, this may not work. OTOH I still have not figured out even an idea to do HFT without colocation, and even then I really miss a proper trading idea.
If you have slower algorithms, VMS have a LOT of advantages, like system stability (no driver updates in the more complex OS, only on the hyper-visor). I am pretty much on a VM for everything, including VOIP stuff. Works like a charm. The ability to easily rescale requirements is NICE.
One of them virtualization heavy 
