Hi Adam,

I've set the max clients higher than your requirement without issue for many years.

it's a good idea to also set the following parameters in the dacs.env for usage capture when using a remote dacs call.

DACS_SNKD_P2P_USAGE

DACS_SNKD_P2P_USAGE_SIZE

I would recommend you get your network team to put a GTM style load balancer address in front of the dacs_snkd's so you at least have a failover in the event of a single dacs_snkd failing and useful to distribute the load across a few dacs_snkd's.

I've used a high client number for many years starting back in the days of Kobra/SASS3 RV mode running over SASS3 and where we did not want local dacs_snkd's on every client PC and used a farm of dacs_snkd's sitting on servers.

Regards Paul

Hello @adam.martin,

Refinitiv developers forums are dedicated to general questions and discussions centered around Refinitiv API usage. The majority of forum's members are Refinitiv API developers, moderators are Refinitiv API experts. These are the questions that this forum is most helpful with.

In my view (developer), as a single dacs.snkd is required to be fast, it has gone through release and testing under max of 8 connections, I would not be comfortable increasing max connections to a number far beyond the recommended. As my understanding is, that you are designing for production use, to me, memory requirement would not be the only concern, a full system test under this configuration setting would be required to have reasonable confidence.

If the targeted daemons run on DACS, is remote to the client app, your market data team/admin, who manage your DACS, and the daemon, would be the ultimate decision maker on any config changes, so they would be the first to consult, and if the are a "no", it's a "no-go". If it's your local daemons, you can likely increase the parameter yourself, but then the responsibility for doing so, and the implied system testing would be on you.

You may, as a customer, raise a case with Refinitiv DACS support, via Refinitiv Helpdesk Online perhaps DACS support engineers have more in-depth guidance to share on this matter.

Would it make sense to discuss your use case, why 300-400 connections per daemon are required, perhaps we can be of help in that?