# Self-run SASE suits teams that can staff policy changes and incidents, and a managed model suits teams that cannot

Guide, 12 Nov 2026, read 12/13 Nov 2026

**The short answer.** Decide managed or self-run by who answers the pager and who approves policy changes. Aryaka offers self-managed, co-managed and Aryaka-managed choices. Open Systems sells three operating models on one platform. AT&T announced a fully managed service on Palo Alto Networks on 29 Jul 2026. Standard support at iboss is live 8am to 8pm ET on weekdays.

## The question is staffing, not features

Every SASE platform can be run by the customer. The question is whether you will. A policy change at a branch, a certificate problem on TLS inspection, a fault between a branch and a PoP: someone has to own each, on every shift your sites operate. If your team has two network engineers and 150 sites, the platform is not the constraint.

I have seen self-run work well in a team of six with a clear change process, and fail in a team of twelve with none. The headcount is less important than whether the change process exists.

Write the staffing number down. Count the hours per month your team spends today on policy requests, certificate and agent problems and incident triage, then add the migration load for the first year. If that exceeds one engineer's time and you have no spare engineer, a managed or co-managed model is cheaper than hiring.

## Operating models as published

This table lists what the vendors and carriers publish about who runs the service. All entries are their statements.

| Provider | Model as published | Detail |
| --- | --- | --- |
| Aryaka | Self-managed, co-managed or Aryaka-managed | Most customers live in under 4 months, with implementation, operations and 24/7 support included as a service (Aryaka states) |
| Open Systems | Self-serve Platform, AIOps or Mission Control | In Mission Control, Level-3 engineers own every critical call, with no Level 1, Level 2 or queue (vendor) |
| Cato Networks | Managed SASE through MSP partners | SMB FlexPool, a pooled-license program for MSPs, announced 1 Sep 2026 |
| Versa | Titan for lean IT; sales 100% through partners | Versa NOC covers the cloud infrastructure round the clock; MSP multi-tenancy and billing |
| Palo Alto Networks via AT&T | AT&T SASE as a fully managed service, announced 29 Jul 2026 | AT&T manages deployment, policy and operations (press release) |
| Cisco via AT&T | AT&T SASE with Cisco, announced 27 Aug 2025 | Built on Secure Access with SD-WAN |
| Arista VeloCloud | Service providers run the platform | T&A SYSTEME manages 7,500+ sites in 98 countries (Arista release) |
| iboss | Direct support or MSP platform | Standard support 8am to 8pm ET weekdays; 24x7 live support in a higher package |

## What a managed model takes off your plate

In the fully managed models the provider handles deployment, policy changes and incident response. Palo Alto Networks and AT&T say AT&T manages deployment, policy and operations. NTT DATA and Palo Alto Networks signed a multi-year alliance on 20 Aug 2026 that includes Zero Trust and SASE managed services, targeting $1 billion in joint business by 2029. That is a sales target, not a service level, but it shows where the channel is investing.

What you get: a named operator, an SLA on change requests if you negotiate one, and no hiring plan. What you give up: direct admin speed, and some visibility into why a decision was made.

## What stays with you either way

Application owners still have to say who needs access to what. The provider cannot decide that. Incident triage between the SASE service and your application teams is still a human conversation. And someone on your side still needs read access to logs, because the first question in any dispute is whether the provider's data matches yours.

Aryaka's third-party circuits are an example: repair time and availability follow the ISP, and Aryaka states it is not responsible for an ISP failing to restore service, even under a managed model.

If you pick a carrier-managed service, compare its change-request terms to the vendor's direct terms, since the carrier adds its own queue on top of the vendor's support.

## Support hours are a hidden contract term

Support terms vary within one product line. iboss states standard support is live 8am to 8pm ET on weekdays, with 24x7 live support in the higher package, for customers who buy directly. A company with sites in three time zones needs the higher package in the order form.

Check what response time applies to a policy change request, not just to an outage. Many contracts define only the latter.

## Co-managed in practice

Co-managed is where I end up most often: the provider runs the platform and the first line, and you own policy and application decisions. It works when the split is written down. Put in the contract who may change what, with what lead time, and how emergency changes are logged. If the provider also holds your only admin accounts, you cannot leave without their cooperation, so keep at least two break-glass admins on your side.

## Exit and change of model

Open Systems says customers can move between its three operating models without changing platform or provider. That is a useful property if true, since needs change. Ask any provider the equivalent: can we take admin back, and what do we get in a handover? Write the answer into the order form while you can still negotiate.

Also ask how the managed fee scales. A fee per user or per site is predictable. A fee based on tickets or change counts is not, and it punishes you in the busy first year.

## Questions that decide it

Five questions settle most cases.

- Who answers a branch outage at 2 a.m. in each time zone you operate in?
- How long does a policy change take from request to live, and who approves it?
- Do you have two people who know the platform well enough to cover vacations?
- Does the provider give you read access to raw logs and configuration history?
- What is the exit path: admin handover, data export, notice period?

## Sources

- Aryaka, Unified SASE platform: https://www.aryaka.com/unified-sase-platform/ (read 12 Nov 2026)
- Aryaka, SASE reviews: https://www.aryaka.com/sase-reviews/ (read 12 Nov 2026)
- Aryaka, services terms: https://www.aryaka.com/services-terms/ (read 12 Nov 2026)
- Open Systems, home page: https://www.open-systems.com/ (read 12 Nov 2026)
- Open Systems, AI press release: https://www.open-systems.com/blog/press-release/ai-just-reset-the-race/ (read 12 Nov 2026)
- Cato Networks, SMB FlexPool: https://www.catonetworks.com/news/cato-launches-smb-flexpool/ (read 12 Nov 2026)
- Versa, lean IT: https://versa-networks.com/solutions/lean-it/ (read 12 Nov 2026)
- Palo Alto Networks, AT&T press release: https://www.paloaltonetworks.com/company/press/2026/at-t-business-and-palo-alto-networks-tackle-the-growing-complexity-of-securing-distributed-enterprises (read 12 Nov 2026)
- Palo Alto Networks, NTT DATA alliance: https://www.paloaltonetworks.com/company/press/2026/ntt-data-and-palo-alto-networks-sign-global-strategic-alliance-to-accelerate-secure-ai-transformation (read 12 Nov 2026)
- Cisco, AT&T SASE with Cisco: https://newsroom.cisco.com/c/r/newsroom/en/us/a/y2025/m08/at-t-and-cisco-deliver-comprehensive-secure-access-service-edge-sase-solution.html (read 12 Nov 2026)
- Arista, ETM for VeloCloud SD-WAN press release: https://www.arista.com/en/company/news/press-release/24357-pr-20260721 (read 12 Nov 2026)
- iboss, service level agreement: https://www.iboss.com/service-level-agreement (read 12 Nov 2026)

Last reviewed 12 Nov 2026. Canonical: https://bestsasesolutions.com/guides/managed-vs-self-run-sase/
