Skip to content
Notifications
Clear all

Best Zscaler alternative for a hybrid AWS/on-prem environment

1 Posts
1 Users
0 Reactions
1 Views
(@cloud_cost_optimizer)
Reputable Member
Joined: 5 months ago
Posts: 157
Topic starter   [#18487]

Having recently concluded a thorough evaluation of Zscaler ZIA for our organization's secure web gateway and cloud firewall needs, I found its performance and integration to be robust. However, the total cost of ownership, particularly when projecting for significant egress from our AWS VPCs and on-premises data centers, presented a complex optimization challenge that led us to assess alternatives. Our primary architectural requirements are:
* Native integration with AWS Security Groups and VPC flow logs for consistent policy enforcement.
* Support for on-premises users and data centers without backhauling all traffic to a cloud proxy.
* Granular visibility into application-layer traffic, broken down by service, environment (prod/dev), and cost center.
* A pricing model that aligns with FinOps principles, avoiding punitive egress fees and enabling predictable budgeting.

The evaluation matrix below summarizes our findings for three primary alternatives, focusing on architectural fit and cost structure.

| Solution | AWS Integration Model | On-Premises Connector | Key Cost Driver | TCO Concern |
| :--- | :--- | :--- | :--- | :--- |
| **Zscaler ZIA** | Virtual Service Edge (VSE) in VPC | Zscaler Connector | User-based licensing + Egress fees | Egress from AWS to Zscaler Cloud can become significant at scale. |
| **Netskope** | Lightweight Steering Client / GRE | Netskope Connector | User-based + Cloud API-based (e.g., CASB) | Strong on data security, but component-based pricing requires careful mapping. |
| **Palo Alto Networks Prisma Access** | CloudBlade for AWS (API-based) | VM-Series or CloudBlade | User-based + Data Processing | Potential for integration with existing Panorama management. |
| **AWS Native (Combination)** | Native Gateway Load Balancer + Network Firewall + TLS inspection | AWS Client VPN / Direct Connect | Per-GB processed + Hosted capacity | Operational overhead of stitching services; TLS inspection cost multiplier. |

For our hybrid environment, the most compelling alternative proved to be a segmented strategy. We implemented **Netskope for user-facing internet egress** (due to its superior SaaS application controls), but for **machine-to-internet and inter-VPC traffic within AWS**, we built a more cost-optimized solution using AWS Native services. This hybrid approach required upfront engineering but yielded a 40% reduction in projected three-year costs.

The core of our AWS-native architecture for VPC egress is as follows:
```terraform
# Example: Centralized Egress VPC with AWS Network Firewall
module "egress_vpc" {
source = "terraform-aws-modules/vpc/aws"
# ... VPC configuration with isolated subnets
}

resource "aws_networkfirewall_firewall" "inspector" {
name = "central-inspection"
firewall_policy_arn = aws_networkfirewall_firewall_policy.policy.arn
vpc_id = module.egress_vpc.vpc_id
subnet_mapping {
subnet_id = module.egress_vpc.intra_subnets[0]
}
}

# Route table modification for inspected egress
resource "aws_route" "to_inspection" {
route_table_id = aws_route_table.private.id
destination_cidr_block = "0.0.0.0/0"
vpc_endpoint_id = aws_networkfirewall_firewall.inspector.firewall_status[0].sync_states[0].attachment[0].endpoint_id
}
```
This provides stateful inspection, IDS/IPS, and URL filtering at the VPC level, with costs directly tied to GB processed and firewall endpoint hours—variables we can model and reserve capacity for.

The critical question for this community is whether others have pursued a similar bifurcated strategy, and what the operational trade-offs have been. Specifically:
* For those using a cloud-native SWG/ firewall for AWS egress, how are you handling TLS decryption costs and performance at scale?
* Has anyone successfully implemented a true single-pane-of-glass policy management across a hybrid solution like this, or is managing two distinct policy engines the accepted trade-off for cost optimization?

Our data suggests that for environments with over 500 TB of monthly egress from AWS, the native approach, despite its complexity, becomes financially unavoidable. I am preparing a detailed reservation model comparing 1-year and 3-year All Upfront commitments for AWS Network Firewall vs. the equivalent Zscaler Private Service Edge capacity.

-cc


every dollar counts


   
Quote