Skip to main content
AI Data Engine

Known limitations in AI Data Engine

Contributors netapp-dbagwell
PDFs

Known limitations identify functionality that isn't supported in AI Data Engine (AIDE), or functionality that doesn't interoperate correctly. Review these limitations before deploying or using the product.

Deployment

Single-node and multi-node deployments are independent

An AIDE single-node deployment and an AIDE multi-node deployment are separate deployment types. There is no supported migration path to move an existing deployment from one type to the other.

Proxy support is limited

You can deploy AI Data Engine with a forward proxy, but the following proxy features aren't supported:

  • NTLM, Kerberos, or client-certificate authentication to the proxy. Use no authentication or basic authentication.

  • SOCKS proxies.

  • Proxy auto-configuration (PAC file or WPAD). Give the installer the proxy host and port directly.

  • A different proxy for different destinations. One proxy carries all AI Data Engine internet traffic.

  • An IPv6 address as the proxy address. Use a host name instead; the host name can resolve to IPv6.

  • A proxy with an air-gapped installation. The installer rejects proxy options on an air-gapped installation.

    For supported proxy configurations, see Learn about proxies for AI Data Engine.

Policies and actions

Policy-triggered actions are limited

When you create a policy, the available automated actions are limited to permanent deletion and email notification. Automated archive and tag actions from policies are not available. You can create a policy with no action to monitor a data condition over time.

Tags

Tags can't be deleted or renamed

After you create a tag, it remains available in the system. You can apply a tag to items and remove it from items, but you can't delete the tag definition itself, rename it, or correct a tag name that contains a typo. Tag names are case-sensitive, so Finance and finance are different tags.

Governed AI access (MCP)

MCP search filters are limited

AI clients can search files by name prefix, file type, and owner only. If you ask for files by size, date, or folder, the AI client must page through results, and the answer might be incomplete.

MCP tag operations are limited

AI clients can't list the files that have a tag, and they can add or remove a tag on only one file at a time.