Skip to main content

Where your process data lives is your decision

Novasign Studio runs on your own servers, in your own cloud account or on an appliance that Novasign builds and ships, and Managed Novasign Cloud, which Novasign runs for you, is rolling out. The on-premise and appliance options can run air-gapped, and you decide where your process data is stored. The same installation grows with you, from one scientist to a whole department and from a single model in training to many trained side by side.

Book a demo
On-premiseAvailable TodayYour cloudAvailable TodayApplianceAvailable TodayManagedNovasign CloudRolling OutSite, Power, NetworkServersClusterStudio and MaintenanceYouYouNovasignNovasignYouYouNovasignNovasignYouNovasignNovasignNovasignNovasignNovasignNovasignNovasignRun by NovasignSign In Through Your Identity Provider in All Four Options
  1. On-premise (Available Today). You: site, power, network, servers. Novasign: cluster, studio and maintenance.
  2. Your cloud (Available Today). You: site, power, network, servers. Novasign: cluster, studio and maintenance.
  3. Appliance (Available Today). You: site, power, network. Novasign: servers, cluster, studio and maintenance.
  4. Managed Novasign Cloud (Rolling Out). Novasign: site, power, network, servers, cluster, studio and maintenance. Run by Novasign.
  5. Sign In Through Your Identity Provider in All Four Options
  • On-premiseAvailable Today

    You provide the servers and the network segment. Novasign provides and maintains the platform.

  • Your cloudAvailable Today

    Studio runs in your own cloud account and region, so your data stays in the region you choose.

  • ApplianceAvailable Today

    Novasign builds, tests and ships a dedicated server. You provide rack space, power and a network connection.

  • Managed Novasign CloudRolling Out

    Novasign runs the environment for you from end to end. It is rolling out, and no general availability date is published yet.

Setting up a deployment

Where it runs and what may leave it are your decisions, and sign-in connects to the identity provider you already use.

  1. 1

    Choose where it runs

    Novasign sizes the deployment with your team. Each node needs at least 24 vCPUs and 64 GB of RAM, on a private network segment with no public ingress. For a proof of concept, Novasign can host Studio for you during a trial period, so you can evaluate it before setting up any infrastructure.

  2. 2

    Connect your identity provider

    Studio signs people in with SAML 2.0 or OpenID Connect against your identity provider, such as Microsoft Entra ID, Okta or AD FS, or against LDAP or Active Directory, and can provision accounts with SCIM 2.0. Your provider’s multi-factor policy applies to those sign-ins. Accounts outside it can use one-time codes or passkeys, which each organization can make mandatory.

  3. 3

    Decide what may leave

    On-premise and appliance deployments store your projects, models, datasets and process data inside your network, and an air-gapped installation sends nothing outside it. Where a deployment opts into connected monitoring, you review its scope with your security team first. Credentials for the systems Studio connects to are resolved at run time and never sit in a workflow or in the browser.

Studio scales with the work you give it

Novasign Studio is built for one user or a whole department, for a small dataset or a very large one, and for a simple model or an integrated process model that needs far more computation. It distributes runs across a pool of workers that run side by side, and each worker uses the cores and memory it is given.

Scaling Across Your Cluster
Novasign StudioYour ClusterOne ResultYour WorkflowWorkerCores · MemoryWorkerCores · MemoryWorkerCores · Memory+ A Worker You AddGathered in Order
  1. First, your workflow in Novasign Studio, shown as three connected nodes on a card that carries the Novasign logo.
  2. Its work fans out into a dashed region labelled as your cluster.
  3. Inside it, three workers run tasks side by side, each one made of cores and memory.
  4. A dashed empty row stands for a worker you add to scale out.
  5. Last, the results gather back into one set.
  • Horizontal scaling

    Add workers to run more at once

    Capacity is a pool of workers rather than a single machine. Add workers, and Studio distributes the simulations you queue across more of them and gathers the results into one set.

  • Vertical scaling

    Give each worker more cores and memory

    The cores and memory each worker receives are set in the cluster configuration. Raising them leaves every workflow, dataset and model exactly as it was.

  • Data residency

    Scaling stays inside your environment

    Both are settings in the cluster configuration, agreed with your team. The platform, the sign-in, the projects and the private network segment stay the same.

The engines powering Novasign Studio

You draw a workflow, and the kind of distributed engine that large data teams rely on does the heavy work underneath it. All of it ships with the deployment and runs inside it, on the deployment’s own storage. Each engine is a default rather than a hard dependency.

What runs the heavy work
Queries
A distributed query engine answers SQL across converted datasets, so a question about a very large table runs across the cluster instead of on one machine.
Conversions
A distributed compute engine does the conversions and exports, splitting a big table across the cluster rather than pulling it through one process.
Storage
Datasets are held in an open columnar table format on S3-compatible storage inside your own deployment. Your own tools can read them, and no proprietary file format locks them in.
Training
Models are trained on open deep-learning frameworks that the machine-learning research community uses, not on a proprietary one.
Distribution
A distributed executor hands the pieces of a run to the workers and brings the finished pieces back.

Novasign Studio security by deployment option

Six questions an IT security assessment usually asks first, answered for each of the four deployment options.

Where process data is stored

On-premise
On your servers
Your cloud
In your cloud account and region
Appliance
On the appliance in your rack
Managed Novasign Cloud
In a Novasign environment in an agreed region

Public internet exposure

On-premise
None by default
Your cloud
None by default
Appliance
None by default
Managed Novasign Cloud
Agreed with you

Sign-in

On-premise
Your identity provider over SAML 2.0 or OpenID Connect
Your cloud
Your identity provider over SAML 2.0 or OpenID Connect
Appliance
Your identity provider over SAML 2.0 or OpenID Connect
Managed Novasign Cloud
Your identity provider over SAML 2.0 or OpenID Connect

Novasign remote access

On-premise
Scoped, via your jump host or VPN, under named accounts
Your cloud
Scoped, via your jump host or VPN, under named accounts
Appliance
Scoped, via your jump host or VPN, under named accounts
Managed Novasign Cloud
None into your network

Air-gapped operation

On-premise
Supported, engineered per engagement
Your cloud
n/a
Appliance
Supported, engineered per engagement
Managed Novasign Cloud
n/a

Backup

On-premise
Your backup tooling, or hypervisor snapshots
Your cloud
Volume snapshots or your cloud backup tooling
Appliance
RAID plus a backup target you nominate
Managed Novasign Cloud
Operated by Novasign to objectives agreed with you

Each row is answered for all four options. "n/a" means the option does not offer it.

The compliance program

Novasign runs a formal information security and compliance program. It is working towards ISO 27001 and SOC 2 and covers GDPR and the EU AI Act. The Novasign Trust Center documents the program as its evidence is collected, and Novasign answers security questionnaires on request.

Tell us your IT requirements

Every IT organization sets its own policies, for outbound traffic, for identity and for maintenance windows. Share yours, and our team will match them to the deployment option that fits.

Talk to our teamBook a demo