Support Board
Date/Time: Wed, 27 Nov 2024 00:28:18 +0000
Post From: New Teton Order Routing Service for CME, CBOT, NYMEX, COMEX, FairX, upcoming EUREX
[2021-12-23 16:06:45] |
Sierra Chart Engineering - Posts: 104368 |
We were given these responses from a customer of AMP and would like to respond. From AMP: Thank you for your full response, please ask Sierra for their back up plan.....if something does happen. then let us know if your opinion remains the same. Please let us know what they say....very curious what they are telling their users. We have full backup plans. Depends upon what the reason for the failure is. If the CME primary server has an issue the Sierra Chart order routing process will connect to the CME backup server. We have dual connections to the CME. 10 Gb connections. Failover is completely automatic. And we have two sets of these. And eventually will add three sets of these. This is a total of 6 connections. That is an incredible amount of redundancy. All operating on a high-quality network used by high-frequency traders. We have three Internet connections. We have redundant power. We have the ability to easily failover to another server maintaining a real-time copy of all order routing information, in the case of a server failure. We need to know what is meant by "if something does happen" to provide a specific answer. This is just too vague and generalized. From AMP: Order Routing is a very complex operation... We will maintain communication with Sierra and send an announcement once this service has matured. For example, CQG, Rithmic, and TT....all have been performing order routing services for 20+ years...and they still get issues. So please do not be in a hurry... Order routing is actually not a complex operation. It consists of software, network connectivity, and computer hardware (a server). Certainly the software for order routing which is Sierra Chart itself, running in a multiuser mode, with connectivity to the exchange is complex and has a lot of functionality. However, once all of this is developed and tested and proven which it has been, then everything is running fine and automatic. No intervention is needed. There really is nothing new for the most part and what is new, has already been tested and based upon existing functionality. For example, the functionality to import an end of day account balance file from the clearing firm and update all the account balances, is new. This is developed, very straightforward and simple and works reliably day after day. The networking functionality is provided by our infrastructure provider, which provides fully redundant connectivity to the exchanges. They do this for high-frequency trading firms as well .. We are running on an HFT (High-frequency trading) network, where data moves across that network in 2 µs. The hardware, are new Dell R640 machines. These are equipped with multiple drives, including NVME drives. We use a RAID 1 configuration for storage. They have redundant power supplies and have independent power sources in the Aurora data center. We will not have power issues. A real-time copy of everything is mirrored to another data center in Chicago in the Equinix Cermak data center. So if there is a hardware failure, we can failover in a matter of minutes. Although most likely it would take longer than that because we would have to examine exactly the reason for the failure make an appropriate decision as to what to do. A failover may not be the right action to take. We may have to discuss the issue, with our infrastructure provider or the CME GCC. A scenario like this is extremely unlikely and in all of our 15 years of experience we have never suffered a catastrophic hardware failure requiring a failover to another server. The most likely failures relate to Internet and power. For Internet we have triple redundancy which is fully automatic. And we have true redundant power in the Aurora data center. We have a simple monitoring process, for all of our servers and processes running on those servers and we are notified in about one minute, when there is process which is down. So we are able to react to these very quickly. All of our CME processes, do not go down. All of this is highly reliable and we have a track record of nearly 100% uptime with our Denali Data Feed. We only had two incidents, Sunday evening. One of them was with CBOT data due to an empty configuration file from the CME. And another issue related to historical data. Both of these were promptly taken care of. Neither of these issues would affect order routing. Order routing is actually in some ways is more simplified than handling market data. Sierra Chart has provided order routing services for 15+ years. And we have also used a centralized model for about six years. So we are also highly experienced. And we have less issues, than TT, and CQG and Rithmic. For example with our TT based order routing, we have had only one issue on our side lasting a few hours, and it overwhelmingly was caused by a TT related issue. Otherwise, it has been a perfect track record, aside from the issues from TT. How often does anyone have trouble logging into Sierra Chart due to an issue on the Sierra Chart server-side. It does not happen. It never happens. Although if you do not save your password within Sierra Chart and you try to login and we have an issue with our Web server at that exact moment, then yes you would have a problem. Otherwise, there is a nonissue due to how everything is designed. We do have redundancy with our Web server as well. Furthermore, logging into our order routing servers, it is something that almost would never encounter a failure as long as you are authorized to connect because of how that is designed. They say "do not be in a hurry". No problem. Customers can make their own decisions and decide to move their accounts where they want. We are not in any hurry. We have worked on CME order routing, for about two years and have it all perfected and stable. Most of the time was related to sequence number handling. And also getting through the CME certification (very frustrating). We did not rush anything along. Sierra Chart Support - Engineering Level Your definitive source for support. Other responses are from users. Try to keep your questions brief and to the point. Be aware of support policy: https://www.sierrachart.com/index.php?l=PostingInformation.php#GeneralInformation For the most reliable, advanced, and zero cost futures order routing, *change* to the Teton service: Sierra Chart Teton Futures Order Routing Date Time Of Last Edit: 2021-12-29 13:01:32
|