When a CRM polls VICIdial once per agent: the load you don't see
A CRM that asks VICIdial about each agent separately can quietly become your server's busiest user. How to spot it and the one-call fix.
Who is the busiest user on your VICIdial server? Sometimes it isn't an agent or a manager. It's a CRM that asks about every agent, one at a time, all day, and never shows up in a single VICIdial report.
If agent screens lag and pages load slowly while the floor hasn't grown, this is worth checking before you buy hardware.
What once-per-agent polling looks like
Most CRM integrations want two things from VICIdial: who is logged in, and what each agent is doing right now. With that, the CRM can open the right customer record or show a status light next to each name.
The quick way to build it is a loop. Every few seconds, for every agent, the CRM sends one request to VICIdial's Non-agent API using the agent_status function, which returns the real-time status of one agent. With ten agents, nobody notices. With sixty, that is sixty requests every round, whether anything changed or not.
Each request lands on the web server, checks the API user's login and reads the database. None of them is heavy alone. Together they make a steady stream of work that VICIdial doesn't report, because to VICIdial it's just traffic.
flowchart LR
CRM[CRM loop one request per agent] --> WEB[VICIdial web server]
SCR[Agent screens asking every second] --> WEB
WEB --> DB[(Database)]
DIAL[Dialers placing calls] --> DB
DB -.->|when busy| WAIT[Screens and saves wait]How big it gets on a real floor
On one floor we measured, a CRM polling the API once per agent made up 25–39% of all web requests to VICIdial. That's a quarter or more of everything the web server did, spent on questions whose answers had barely changed since the last round.
That stream competes with traffic that matters. The agent screen already asks the server for news every second, and every Disposition an agent saves has to reach the database too. In our layout, the same machine serves the agent screens and the API and runs the database, so the extra load lands in one place.
When that machine backs up, requests queue. The first thing your agents notice is a screen that stops updating; our checklist for a frozen agent screen covers the other usual causes.
The fix: ask once, for everyone, and keep the answer
The CRM doesn't need sixty answers. It needs one list. The Non-Agent API's logged_in_agents function returns every logged-in agent in one response, with each agent's campaign, status and current lead, and you can narrow it to certain campaigns or user groups. Our walkthrough of logged_in_agents shows the request and what comes back.
The fix we apply has three parts:
- One call for all logged-in agents, instead of one call per agent.
- Cache the answer on the CRM's side, so every CRM page and every user reads the same copy instead of each asking VICIdial.
- Poll less often. Ask how fresh the status really needs to be. A status light that's a few seconds old is fine; a request per agent per second isn't.
How to spot it on your own server
You won't find a CRM line in VICIdial's reports. Look in the web server's request log instead: the same API request repeated with a different agent each time, at a steady rhythm, from one address. If the integration fills in the API's source field, it tells you exactly which system is calling.
Then compare that stream with everything else. If one integration is a big share of what the web server answers, fix it first. A bigger server just answers the same pointless questions faster.
Why it matters more on a cluster
On a cluster, the database and web server is the machine every dialer and every agent screen depends on, so wasted requests there slow the whole floor. Our walk-through of how a VICIdial cluster is laid out explains how that machine fits with the dialers and the standby database.
If one server still carries your floor once the CRM is fixed, a single-server plan may be all you need. If you've outgrown it, book a sizing call and see today's cluster prices.
Frequently asked
- They'll be a little older, not wrong. One shared answer refreshed on a sensible rhythm is plenty for showing who is talking, paused or waiting.
- No. The fix lives in how the CRM asks, and it works the same on one server or a cluster. Fix it first, then size the hardware.
› Will polling less often make the CRM's status lights wrong?
› Do I need a cluster to fix this?
Related from VICIfast
- Running VICIdial across several serversA database, dialers and a standby, sized from load we measured on a live floor.
- Dedicated VICIdial hardware, live pricesEvery core yours, mirrored NVMe, delivered in minutes to days. Priced from live stock.
- When you need dedicated coresAMD EPYC dedicated cores for steady volume or visible CPU steal.
About VICIfast
We run VICIdial servers for call centres, so you don’t have to. You get the dialer set up, secured and kept running — we handle the server, the updates and the backups. Bring your own phone carrier at no markup, or use VICIfast Voice, our own US calling and numbers, from $0.0085 a minute. From $49 a month per server.
Questions? Call +1 636 556 0022 and talk to someone who runs dialers.
Citing this article
VICIfast Engineering. “When a CRM polls VICIdial once per agent: the load you don't see”. VICIfast LLC, November 22, 2026. Retrieved from https://vicifast.com/blog/vicidial-crm-polling-api-load
Have questions?
Related posts
You might be interested in
Guides & tutorials
Idle campaigns left active: the hidden database load in VICIdial
Guides & tutorials
Why searching a lead by phone number freezes VICIdial, and the missing index

Guides & tutorials
Keep the call log small: log archiving on a busy VICIdial database

Guides & tutorials
Reports and tools that slow a live VICIdial floor
VICIfast newsletter
Liked this? Get the next one in your inbox.
We ship the kind of stuff you just read - concrete, numbers-first, no drip. One email when a new post goes live. Unsubscribe in one click.
Comments
No comments yet - be the first.



