in each Region ● AZ: One or more Availability Zones in each DC ○ Independent power, cooling, networking, etc. ● PoP: 20+ Points of Presence, locations across globe peering with the Internet closer gateways: - apiVersion: networking.istio.io/v1beta1 kind: Gateway spec: ... virtualServices: - apiVersion: networking.istio.io/v1beta1 kind: VirtualService destinationRules: - apiVersion: networking.istio.io/v1beta1 kind: DestinationRule spec: ... ... serviceEntries: - apiVersion: networking.istio.io/v1beta1

w/o requiring intermediate Gateway ■ Multiple networks ● all goes though the Gateway ● via L3 networking (if enhanced performance is desired) #IstioCon Demo #IstioCon Istio VM integration seems closer multi-vendor services ○ End-to-end security! (not just between middle boxes) ● High performance networking ○ Much higher multi-Gbps peak data speeds ○ Ultra low latency ○ And of course, reduce overheads first-class support for VM Multiple Networks ○ All traffic goes though the Gateway ○ Need to setup L3 networking if enhanced performance is desired ● Overheads introduced ● No high performance data path support

service mesh enabled ● Reference Agenda #IstioCon Knative and Istio Istio is the default networking layer solution of Knative. It is leveraged for Net-istio is A Knative ingress controller for blue/green deployment How Istio is leveraged in a Knative based platform 90% 10% apiVersion: networking.istio.io/v1beta1 kind: VirtualService spec: gateways: - knative-serving/knative-ingress-gateway vCPU 2 vCPU 2 Gi 4 Gi Y, min=3, max=20 Istiod 1 vCPU 1 vCPU 2 Gi 4 Gi Y, min=3, max=6 Knative Networking-istio 30m 80Mi 900m 2 Gi N #IstioCon Istio scalability optimization during Knative Service provisioning

Secure by default ○ Secret Discovery Service (SDS) ○ Auto mTLS ● API and feature promotion ○ Networking/Security APIs ○ Virtual Machine expansion/Multi cluster mesh https://istio.io/latest/blog/2

All rights reserved. Key Platform Requirements Multi-Vendor Real-Time (RAN) Workload Mobility Networking outside CNF Encryption & Authorization between CNFs 5 ©2021 Aspen Mesh. All rights reserved

make_hazard_pointer_arrray(), e.g., 4, 5, 6 ns vs 4, 8, 12 ns • Optional dedicated thread pool for asynchronous reclamation: • Robustness against latency spikes and deadlock. • Domains: • Robust default domain 24 © 2021. 25 Hazard Pointers Proposal for C++26 Minimalist useful subset of TS2: • Supports asynchronous reclamation • Compatible with external link counting and automatic retirement • Strict subset

Gallery application • Paint Looking at each element outlined in the table in greater detail: Networking, Web browsing and E-mail. Network setup on both Ubuntu and Microsoft Windows is easy. Web browsing

thread. Connection pooling is very, very strongly recommended. • hsha Half synchronous, half asynchronous. The rpc thread pool is used to manage requests, but the threads are multiplexed across the different