Zscaler 블로그

Zscaler 블로그의 최신 업데이트를 이메일로 받아보세요

Best Practices

How SASE Vendors Design Their Solution - Part 2: The Peering Truth—What Does a Good Strategy Look Like?

JOE ARRA, JULIAN WEINBERGER
November 13, 2026 - 5 분 소요
>How SASE Vendors Design Their Solution - Part 2: The Peering Truth—What Does a Good Strategy Look Like?

The Vanity Metric Trap: Why More POPs Doesn't Mean Faster Traffic

In the SASE marketing playbook, the most pervasive vanity metric is point-of-presence (POP) count. Providers frequently advertise hundreds of global locations to suggest ubiquitous proximity. However, field engineers know that routing physics tell a very different story.

Many alternative SASE architectures achieve inflated numbers by deploying dozens of shallow, virtualized mini-POPs. These edge nodes are frequently single-homed to a local ISP with zero direct peering to critical SaaS applications. When users connect, the node must backhaul traffic across congested public transit to reach primary destinations. Proximity to the user provides zero benefit if the edge lacks high-speed upstream connectivity, leading to packet loss and latency spikes.

Why You Can't Rely on Putting SASE Inside Cloud Service Providers (CSPs)

To avoid physical infrastructure investments, several first-generation cloud security vendors built their entire enforcement tier inside third-party Cloud Service Providers (CSPs) such as AWS, Google Cloud, or Microsoft Azure. While this allows rapid geographical expansion, it introduces severe architectural bottlenecks.

First, CSPs are designed as walled computing environments, not carrier-grade transit exchanges. When enterprise traffic enters a SASE enforcement point hosted in a public cloud, it is subjected to public cloud routing limitations. CSP networks optimize for their own internal service delivery rather than external transit. If a remote worker connects to an AWS-hosted SASE POP to access Microsoft 365 or Salesforce, the traffic must traverse complex routing gateways, hop across public cloud boundaries, and endure latency-inducing "hairpins."

Second, public cloud economics directly impacts network performance. Moving traffic out of public clouds incurs exorbitant egress costs. When vendors pay per gigabyte to send traffic across cloud borders, they face a perverse economic incentive: optimize for the cheapest egress path rather than the fastest path. This financial friction inevitably compromises enterprise throughput.

The Power of Major Strategic Hubs: Dense, Multi-Homed Peering

A mature, high-performance SASE architecture takes the opposite approach by establishing dedicated edge deployments within major carrier-neutral Internet Exchange Points (IXPs) (such as Equinix, CoreSite, and DE-CIX). Rather than scattering low-capacity instances across hundreds of disconnected sites, this model concentrates capacity into robust regional transit hubs.

Operating under dedicated Autonomous System Numbers (ASNs), these major hubs implement a symmetric peering strategy that provides three decisive advantages:

  1. Direct SaaS Cross-Connects: Major IXPs house hyperscalers and key enterprise SaaS applications under one physical roof. Through bilateral peering, the SASE data plane hands off inspected traffic directly to Microsoft 365, Google Workspace, or AWS across local dark-fiber cross-connects in under two milliseconds.
  2. Dense BGP Multi-Homing: Enterprise-grade hubs maintain simultaneous connections to multiple Tier 1 backbones and local eyeball ISPs. Through sophisticated BGP routing policies, the platform detects upstream degradation and dynamically shifts traffic paths without dropping active sessions.
  3. Deterministic Symmetric Routing: Dispersed mini-hubs and CSP environments frequently cause asymmetric routing, where request and return packets traverse disparate paths. Major interconnect hubs maintain deterministic, symmetric paths across the entire transaction lifecycle.

 

Architectural Summary: Peering Strategy Trade-offs

Architectural Dimension

Major Strategic Interconnect Hubs

Dispersed CSP & Mini-POP Deployments

Interconnection Model

Direct private/public peering at carrier-neutral IXPs

Leased compute in CSPs; single-homed mini-POPs

Routing Efficiency

Sub-2ms direct handoff to major SaaS endpoints

Multi-hop routing hairpins across cloud boundaries

Carrier Redundancy

Multi-homed across multiple Tier 1 backbones

Dependent on single ISP transit or CSP gateway policies

Cost & Latency Alignment

Flat-rate dark fiber; routing optimized strictly for speed

Egress fees incentivize lower-cost, higher-latency paths

Ultimately, a world-class SASE architecture is defined by the quality of its interconnections, not the quantity of its marketing dots. Concentrating traffic in carrier-grade hubs with rich peering delivers the predictable, low-latency performance that modern enterprise environments demand.

FAQs

POP count is a poor indicator of SASE performance. What matters is how each POP connects to the rest of the internet. POPs in major carrier-neutral Internet Exchange Points (IXPs), with direct peering to SaaS providers and connections to several Tier 1 carriers, deliver lower and more predictable latency. Two other models perform worse: many small, single-homed "mini-POPs", and enforcement hosted inside public clouds.

Hosting SASE gateways inside third-party Cloud Service Providers (CSPs) like AWS, Google Cloud, or Microsoft Azure introduces severe public cloud routing limitations. CSP networks are designed as walled computing gardens to optimize internal service delivery, not carrier-grade transit exchanges. Consequently, enterprise traffic must navigate multiple routing gateways, cross cloud boundaries, and endure latency-inducing "hairpins" to reach external destinations like Salesforce or Microsoft 365.

Public cloud economics create a perverse economic incentive that can compromise enterprise throughput. Because moving traffic out of third-party public clouds incurs high egress costs per gigabyte, vendors hosted in CSPs are financially incentivized to route traffic along the cheapest egress paths rather than the fastest, lowest-latency paths.

Deploying SASE gateways in major carrier-neutral IXPs (such as Equinix, CoreSite, or DE-CIX) allows vendors to bypass public transit entirely. Under dedicated Autonomous System Numbers (ASNs), these hubs leverage direct SaaS cross-connects to hand off inspected traffic directly to hyperscalers and SaaS endpoints like Google Workspace or Microsoft 365 over local dark-fiber connections in under two milliseconds.

Dense BGP multi-homing ensures that SASE hubs maintain simultaneous, redundant connections to multiple Tier 1 carrier backbones and local eyeball ISPs. Through sophisticated Border Gateway Protocol (BGP) routing policies, the SASE platform can instantly detect upstream network degradation or fiber cuts and dynamically shift traffic paths to a healthier carrier without dropping active enterprise user sessions.

form submtited
읽어주셔서 감사합니다

면책 조항: 이 블로그 게시물은 정보 제공만을 목적으로 Zscaler에서 작성하였으며, 정확성, 완전성 또는 신뢰성에 대한 어떠한 보증도 없이 "있는 그대로" 제공됩니다. Zscaler는 제공된 정보를 기반으로 수행된 조치나 오류 또는 누락에 대해 어떠한 책임도 지지 않습니다. 이 블로그 게시물에 링크된 제3자 웹사이트나 리소스는 편의를 위해서만 제공되며, Zscaler는 해당 웹사이트의 콘텐츠나 관행에 대해 책임을 지지 않습니다. 모든 콘텐츠는 예고 없이 변경될 수 있습니다. 이 블로그에 접속함으로써 귀하는 본 약관에 동의하며 귀하의 필요에 따라 정보를 확인하고 사용할 전적인 책임이 귀하에게 있음을 인정합니다.

Zscaler 블로그의 최신 업데이트를 이메일로 받아보세요

양식을 제출함으로써 귀하는 당사의 개인정보 보호정책에 동의하는 것입니다.