Deployment Options

The simplest way to get started with jambonz is to create an account on our cloud platform. This is a fully managed service that allows you to start building jambonz applications in minutes. You can sign up for a free 3-week trial and convert to a paid plan at any time.

Many customers, however, prefer to self host jambonz on their own cloud or infrastructure for reasons of privacy, security, control, or performance. For those customers interested in self-hosting, please review our paid support plans, where the jambonz team will work with you to design, deploy, turn up and test your system and provide ongoing support and maintenance.

The jambonz project is largely funded by our support services, and our related devops scripts and deployment expertise are made available only to paid support customers.

Self-hosted deployment options

There are four different deployment configurations for a self-hosted system, depending on your expected traffic volumes:

• “jambonz mini” - a single server deployment that can handle 50-75 concurrent calls. This is an entire jambonz system on one server or VM, useful for small deployments or for development and testing.
• ”jambonz medium” - a jambonz cluster built on multiple servers or VMs that can handle up to 3,000 concurrent calls.
• ”jambonz large” - a jambonz cluster built on multiple servers or VMs that can handle an unlimited number of calls.

jambonz mini

The jambonz mini deployment runs the complete jambonz software suite on a single server. This includes quite a few different components: the SBC and feature server call processing elements, databases and monitoring/tracing software, call recording support and more. Because of that the scalability of the system is limited; for scalability we recommend a clusterered environment, either a jambonz medium or large cluster. However, the jambonz mini is a great choice for a development or staging environment, PoCs, or small production loads.

jambonz medium

A jambonz medium cluster is a horizontally scalable deployment that should be used if you project needing to handle up to 3,000 concurrent calls. It would consist of

• 1 (or more) SBCs
• 1 web / monitoring server
• 1 or more feature servers in an horizontally scalable cluster.

Each SBC will be able to handle 1,500 concurrent calls, and each feature server will generally be able to handle between 300-500 concurrent calls, so you will scale those according to your volumes. These numbers are dependent on the specific hardware and network configuration of your deployment, which we will provide guidance for as part of the paid support plan.

jambonz large

A jambonz large system contains multiple horizontally scalable clusters, and can handle any number of concurrent calls. The system will minimalluy consist of:

• 1 or more SBC SIP servers
• 1 or more SBC RTP servers in a horizontally scalable cluster
• 1 or more feature servers in a horizontally scalable cluster
• 1 web server
• 1 monitoring server

The system scales, first by adding additional feature servers, then later as traffic grows by adding additional SBC RTP servers and finally, if needed, additional SBC SIP servers. This reflects the fact that the feature server is central processing focus point of the system and that handling the media is much more processing-intensive than handling then the sip signaling.

Kubernetes

We provide a helm chart for deploying jambonz on Kubernetes for our paid support customers that prefer that. However, there are challenges in running VoIP on Kubernetes and it requires a complex configuration of special node pools with host networking enabled. Furthermore, performance is less on Kubernetes as we can not run the kernel module for the SBC STP element that we can when deploying on VMs or bare metal. Finally, troubleshooting is more difficult on Kubernetes as well because Kubernetes logging elements like Kibana are not well-suited for troubleshooting SIP log files.

For these reasons, if you are undecided about whether to deploy on Kubernetes or VMs, we recommend deploying on VMs using autoscale groups. On most hosting providers you can build reusable machine images (e.g. AMIs on AWS) which provide many of the same benefits of an easily recreatable and manageble deployment.

Hosting Providers

Under our paid support plans, we help customers deploy on their preferred hosting provider. We have experience deploying on AWS, Azure, Google Cloud, OVH, Hetzner, and many others as well as bare metal deployments.

We provide scripts (e.g. Terraform, Cloudformation, packer etc) to automate the deployment process for you and we train your team in how to properly manage jambonz deployments on your chosen provider.