diff --git a/_posts/2023-07-14-evolution-of-a-vpc.md b/_posts/2023-07-14-evolution-of-a-vpc.md index 84b96d6..7348cff 100644 --- a/_posts/2023-07-14-evolution-of-a-vpc.md +++ b/_posts/2023-07-14-evolution-of-a-vpc.md @@ -19,21 +19,31 @@ what we proposed was a simple VPC connecting all remote sites via site-to-site V > It's always essential to consider as a prerequisite whether your customer gateway peer supports the default behaviour of the AWS side of the connection -The below figure illustrates the initial design - ### _Initial Design_ +The below figure illustrates the initial design + +![image]({{ site.baseurl }}/assets/images/chrysos-init.png) + +We could have just used a transit VPC and connected it to the regional data sites and VPCs via site to site VPN via VGW instead of TGW this would have necessitated highly available layer2 VPN appliances to be maintained by the client and a dedicated network engineer to look after it. +Instance based routing gets costly when its tries to scale up and its constrained by the instance bandwith limitations. + +Whereas TGW is a highly available and scalable managed services and it supports upto 50Gbs burst traffic out of the box . + +### _Iteration of the Initial Design_ ![image]({{ site.baseurl }}/assets/images/generic_initial.png) > Transit Gateway was used to give connectivity between VPCs spread across accounts and the site-to-site connections terminated at the TGW instead of a virtual private gateway. -since the remote sites are even further from the home VPC region and they rely on 3G/4G/satellite internet connectivity, we suggested they use AWS global accelerator on the VPNs to route traffic through the nearest AWS point of presence. +Since the remote sites are even further from the home VPC region and they rely on 3G/4G/satellite internet connectivity, we suggested they use AWS global accelerator on the VPNs to route traffic through the nearest AWS point of presence. -with the inclusion of the Global Accelerator, the design looked as shown below +With the inclusion of the Global Accelerator, the design looked as shown below. ### _With the inclusion of GA_ ![image]({{ site.baseurl }}/assets/images/generic_with_ga.png) +AWS Global Accelerator can enhance the performance of site-to-site connectivity by up to 60%, ensuring faster data transfers and reduced latency. + Imagine you're a customer using a cloud service for important tasks. Just like anyone else, you wouldn't want all your important services to rely solely on one cloud provider. That would be risky, as it could lead to a single point of failure. So, to avoid that, you'd prefer to distribute the risks by not depending entirely on one cloud vendor. @@ -48,7 +58,7 @@ At this point in the design, we had successfully ensured high availability (mean ***What if an AWS region or Palo Alto regional construct fails, will it pull down the rest of the services together with it ?*** -simple answer is **_yes_** . it would inevitably pull the rest of it down and how would you circumvent it? +Simple answer is **_yes_** . It would inevitably pull the rest of it down and how would you circumvent it? So, we came to a decision: to make use of AWS regions. This choice greatly enhances the architecture's ability to handle faults and ensures overall stability. In this setup, each region will have its network hub, just like in the initial design, and we'll maintain the same number of Virtual Private Clouds (VPCs) across AWS Accounts. All these network hubs will be interconnected in a mesh-like pattern. Each Site to site connection originating from Palo Alto Prisma will be terminated at each end of the regional network hubs as shown below. @@ -58,12 +68,13 @@ Initially, every site-to-site VPN connection was rolled out to a single network ![image]({{ site.baseurl }}/assets/images/multi-regional-networkhub.png) -then once the rest of the regional network hubs were ready those redundant connections made their way to their intended regional hub as shown below +Then once the rest of the regional network hubs were ready those redundant connections made their way to their intended regional hub as shown below ### _Site to Site terminating at each regional hub_ ![image]({{ site.baseurl }}/assets/images/multiregional-network-high-tolerance-hub.png) +Reliability has been strengthened to withstand both zonal and regional failures effectively. Managed services such as TGW/Site-to-Site VPN, along with Prisma Cloud, significantly enhance availability, ensuring seamless service without disruptions ## Implementation @@ -72,7 +83,7 @@ then once the rest of the regional network hubs were ready those redundant conne > As we were setting up the orchestrator on a brand-new AWS region, we encountered a situation where certain services were not yet accessible. This prevented us from fully utilizing the orchestrator's complete range of features. To work around this, we chose to programmatically exclude those services by making changes to the code. This way, we could still proceed with the implementation without those specific features. -Due to security and cost optimisation considerations, an egress VPC for each regional network hub was introduced to direct all the outbound internet traffic from the regional spoke VPCs to go through it. In addition, interface endpoints with their own private hosted zones were implemented on the same egress VPC to be shared across the network hub. +Due to security and cost optimisation considerations, an egress VPC for each regional network hub was introduced to direct all the outbound internet traffic from the regional spoke VPCs to go through it. In addition, interface endpoints with their own private hosted zones were implemented on the same egress VPC to be shared across the network hub. This keeps the traffic within the AWS network backbone without relying on the egress internet traffic and reduces the costs significantly. ## Conclusion diff --git a/assets/images/chrysos-init.svg b/assets/images/chrysos-init.svg new file mode 100644 index 0000000..0fdabaf --- /dev/null +++ b/assets/images/chrysos-init.svg @@ -0,0 +1,24 @@ + + + + + + + + + + + Internet gatewayVPCVPCSite-to-Site VPNData centerafrica sitecanada siteNAT gatewayau site sydney Site-to-Site VPNSite-to-Site VPNSite-to-Site VPNGlobalAcceleratorGlobalAcceleratorGlobalAcceleratorVPCSite-to-Site VPNSite-to-Site VPNSite-to-Site VPNFrame 1 Dedicated Highly available VPN appliance will enable transit vpc to route traffic using layer 2 routing \ No newline at end of file