<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Rusty Williams on Medium]]></title>
        <description><![CDATA[Stories by Rusty Williams on Medium]]></description>
        <link>https://medium.com/@sunaj10?source=rss-3e67c661019c------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*dmbNkD5D-u45r44go_cf0g.png</url>
            <title>Stories by Rusty Williams on Medium</title>
            <link>https://medium.com/@sunaj10?source=rss-3e67c661019c------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Tue, 17 Nov 2026 20:18:47 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@sunaj10/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Azure Firewall Force Tunneling with Dnat rules: Swatting the nats]]></title>
            <link>https://medium.com/@sunaj10/azure-firewall-force-tunneling-with-dnat-rules-swatting-the-nats-71d7acbdaafa?source=rss-3e67c661019c------2</link>
            <guid isPermaLink="false">https://medium.com/p/71d7acbdaafa</guid>
            <category><![CDATA[azure]]></category>
            <category><![CDATA[azure-networking]]></category>
            <dc:creator><![CDATA[Rusty Williams]]></dc:creator>
            <pubDate>Wed, 11 Nov 2026 21:22:46 GMT</pubDate>
            <atom:updated>2026-11-11T21:22:46.659Z</atom:updated>
            <content:encoded><![CDATA[<p>Hello all you phillumenists! I wanted to take a minute to re-hash/talk more about azure firewall force tunneling. I mention a lot about it <a href="https://medium.com/aws-in-plain-english/azure-public-ips-and-forced-tunnelling-get-ready-for-a-pummeling-fcd5b900960b">here</a> and talk a little about how to make dnat rules work and leave out any azfw management nic info other than to say it won’t help if you’re not doing force tunneling via UDR’s. Let’s go through that a bit now, trading Bean for Sheen.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/577/1*rf38fFiaAMGKki0TW8zbag.jpeg" /></figure><p>As I mentioned before, once you start advertising a default from onprem the azfw public ip kinda becomes useless as it won’t receive any traffic on it since it knows that internet traffic should come from and go to onprem. Ideally you’ll want to have an AzureFirewallManagementSubnet in place and the management nic enabled. You’ll want to put a UDR on that guy sending internet traffic to the internet to stay “supported” and have all the azfw management goodness happily going to the Azure backend. You have this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*45QViEVvXHWoZMY9s9WVfw.png" /></figure><p>But you find out, “Hey, we have a client that needs to access our VM from a dnat rule, but we still want internet traffic going to onprem. You can set that up right?” Of course you setup the dnat rule and nothing works, because you need this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ZmaYlnvf1H6xOqcnIlwZng.png" /></figure><p>On the azfw subnet you’ll need a UDR with all of the outside ip ranges that’ll be coming into the dnat rules (hopefully there’s not a crazy amount). You gotta have this because the Expressroute gateway will be plumbing a default route into everything and this keeps traffic nice and symmetric, because Azure likes that. If you don’t actually want to force tunnel internet to onprem, you can just slap a default 0.0.0.0/0 route on the AzureFirewallSubnet UDR instead of the specific ranges and call it a day.</p><p>That’s all there is for this one, I just wanted to explain a little more about the force tunneling behavior with dnat rules. Thanks for reading through this and check out my Github for more labs and examples here: <a href="https://github.com/quiveringbacon">https://github.com/quiveringbacon</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/352/0*qBRIy9o-osRO_QgO.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=71d7acbdaafa" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Azure Routing 2: UDR 4 U and Me]]></title>
            <link>https://medium.com/@sunaj10/azure-routing-2-udr-4-u-and-me-6ecb75c45b95?source=rss-3e67c661019c------2</link>
            <guid isPermaLink="false">https://medium.com/p/6ecb75c45b95</guid>
            <category><![CDATA[azure]]></category>
            <category><![CDATA[azure-networking]]></category>
            <dc:creator><![CDATA[Rusty Williams]]></dc:creator>
            <pubDate>Wed, 04 Nov 2026 16:27:13 GMT</pubDate>
            <atom:updated>2026-11-04T16:27:13.749Z</atom:updated>
            <content:encoded><![CDATA[<p>Good day all you cruciverbalists! There’s a nifty new feature on the horizon (aka in preview) for vnets and udr’s…the mythical “disablePeeringRoute” property on the route table. Why should you care? Well if you don’t send traffic from onprem through an nva or firewall before going to a spoke vnet, then you probably won’t care at all…You can move along and get some ice cream, for everyone else, let’s talk about the new feature!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/360/1*fE7iv6ZGVsRwqbuO8Xs1UA.gif" /></figure><p>So far, if you’ve ever wanted to send traffic from onprem through a firewall or nva before going to a peered vnet you needed to match the peered vnet range in a udr pointing to your nva/firewall to override the peering route, like so:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*HfYN2VoF-vggqjKPoPmwtg.png" /></figure><p>You have to do it this way because, of course, the most specific routes win and the peering route is more specific. With this new little beauty enabled, you get similar behavior to what you see with private endpoint udr policies, actually a bit more extreme…With this on, the peered vnet routes don’t even show up anymore on the subnet with this table attached! Naturally this will let you slap a large summary route on the route table and not even worry about those pesky ranges anymore. Like this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JiYY3NnQXGIJMJezQH5kwQ.png" /></figure><p>Just to show what it looks like in a VM’s effective routes check this out:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*n5QNGalK5z2uz-ZZkA1CHA.png" /></figure><p>On the left is using the route table without the feature, as you can see the more specific peering routes will always be chosen instead of the generic 10.150.0.0/16. On the right is with the feature enabled, the peered vnet routes are simply gone, leaving no other choice but to use the 10.150.0.0/16.</p><p>This makes things soooo much easier and is a hugely welcome addition, right now it’s on route tables only, so no AVNM support just yet. Also, in the preview the next hop has to be in the same vnet. <a href="https://learn.microsoft.com/en-us/azure/virtual-network/virtual-network-peering-overview?branch=main#disable-peering-route-propagation-preview">Here</a> are the current official limitations.</p><p>Thanks for reading through this super short blurb on a really nifty new feature, hopefully this will help you out. Check out my Github for more labs and examples here: <a href="https://github.com/quiveringbacon">https://github.com/quiveringbacon</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/352/0*nhPfmit-_gj_VVMm.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6ecb75c45b95" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Azure Expressroute with VPN 2: Hiding in plain sight.]]></title>
            <link>https://medium.com/@sunaj10/azure-expressroute-with-vpn-2-hiding-in-plain-sight-ccac9417550a?source=rss-3e67c661019c------2</link>
            <guid isPermaLink="false">https://medium.com/p/ccac9417550a</guid>
            <category><![CDATA[azure]]></category>
            <category><![CDATA[azure-networking]]></category>
            <dc:creator><![CDATA[Rusty Williams]]></dc:creator>
            <pubDate>Wed, 28 Oct 2026 19:58:47 GMT</pubDate>
            <atom:updated>2026-10-28T19:58:47.623Z</atom:updated>
            <content:encoded><![CDATA[<p>Hello there gracious gongoozlers. Let’s have a chat about a scenario you’ll probably never need and have likely never even thought about. That’s right, my last <a href="https://medium.com/@sunaj10/azure-virtual-network-gateways-2-electric-boogaloo-f7460d3742e6">article</a> was super-niche, this one is…Ultra-niche.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/360/1*oIJK8aoyGHqwo3IbALlT2Q.gif" /></figure><p>Imagine you have some reason that you need to be able to have a vpn tunnel from onprem up through your expressroute to your hub vnet (or vhub) and on to some destination device in Azure, but you aren’t allowed to see any of the data and you don’t control either end vpn device…like this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*tTldSEuL7JJs3NVURYeYSw.png" /></figure><p>I don’t know why you need to do this, maybe you merge with another company and are told this is what you have to do. At any rate, here is a possible solution. You have a securehub with the exr gateway and a firewall in it, because your expressroutes all <em>have </em>to terminate in this subscription and can’t be connected anywhere else. Ok, but these peeps are saying you can’t terminate the vpn in that sub, in fact, you can’t terminate it at all. They need to terminate on <em>their </em>device, they already have a vhub in their subscription and want it terminated there. Well…you can’t do it with these restrictions, there’s just no way to connect to the vhub in another vwan without terminating the tunnel because of all <a href="https://medium.com/aws-in-plain-english/azure-vwan-routing-13-multi-vwanning-258758cf1b38">this</a>. But all is not lost!</p><p>You <em>could </em>do this instead:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*XNzmjwzUgkCLcjduLIpu2Q.png" /></figure><p>Yeah that’s right, you make them setup an NVA and connect it to your vhub so that it gets advertised down to onprem. Since you’re not allowed to terminate the tunnel, they can setup the vpn from onprem to the NVA and then setup another vpn from the NVA to their vhub to keep everything nice and invisible. That way it’s all under their control and the NVA routes everything where it needs to go because it will know about onprem from vpn1 and their vhub environment from vpn2. And hey, it’s less work for you.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/889/0*8PiJE7JQcY24S-pK.jpg" /></figure><p>Not saying it’s the best way but it’s <em>a</em> way to get the job done. Thanks for reading through this and check out my Github for more labs and examples here: <a href="https://github.com/quiveringbacon">https://github.com/quiveringbacon</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/352/0*w2TgAljTedTR9s3p.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ccac9417550a" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Azure Firewall Snat Auto-Learning: Route Server to the Rescue]]></title>
            <link>https://medium.com/@sunaj10/azure-firewall-snat-auto-learning-route-server-to-the-rescue-de48a75ddf71?source=rss-3e67c661019c------2</link>
            <guid isPermaLink="false">https://medium.com/p/de48a75ddf71</guid>
            <category><![CDATA[azure]]></category>
            <category><![CDATA[azure-firewall]]></category>
            <dc:creator><![CDATA[Rusty Williams]]></dc:creator>
            <pubDate>Wed, 21 Oct 2026 18:39:10 GMT</pubDate>
            <atom:updated>2026-10-21T18:39:10.748Z</atom:updated>
            <content:encoded><![CDATA[<p>Hello firewall fanatics! Let’s talk about a new feature that just became generally available, the “Auto-learn IP prefixes” feature. This uses route server to learn your private ip ranges rather than you having to either put them in manually or just leave the default RFC1918 ranges set. Let’s have a look at how it’s done.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/360/1*OVBK-UdBLPA2TnRMeBLtpQ.gif" /></figure><p>If you’re using a virtual hub you don’t need to do anything except flip the switch on the policy here:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*JK3rxvYf6YoWnD5-uyTLVA.png" /></figure><p>Since vhubs already have the bigger, meaner brother to route server baked into the vhub, the route service. If you’re using a regular vnet, you’ll need to add a route server to the mix. You might be wondering “Who cares, why would I want to do any of this?” That’s fair, there’s a good chance you don’t need to worry about it. But if you have something like this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jVZQsUloJWCpuOtWMD0aWw.png" /></figure><p>and you don’t want the firewall to snat traffic to that public range, before you’d have to go into the policy and put that range in the “manually specified ranges” but now you can add a route server and just check the box “Auto-learn IP prefixes” and point the firewall to the route server like so:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*tnfhCRi8IXi_buTlJ_AT2A.png" /></figure><p>Once that’s in place the firewall will check for prefixes every 30 minutes or so and you’ll be reveling in no-snat goodness. Keep in mind this only applies to network rules, application rules always snat no matter what you want.</p><p>That about covers it for this new azfw feature, thanks for reading through this and check out my Github for more labs and examples here: <a href="https://github.com/quiveringbacon">https://github.com/quiveringbacon</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/352/0*R1D1qASgeMR1K_ms.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=de48a75ddf71" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Azure Virtual Network Gateways 2: Electric Boogaloo]]></title>
            <link>https://medium.com/@sunaj10/azure-virtual-network-gateways-2-electric-boogaloo-f7460d3742e6?source=rss-3e67c661019c------2</link>
            <guid isPermaLink="false">https://medium.com/p/f7460d3742e6</guid>
            <category><![CDATA[azure-networking]]></category>
            <category><![CDATA[azure]]></category>
            <dc:creator><![CDATA[Rusty Williams]]></dc:creator>
            <pubDate>Mon, 12 Oct 2026 13:04:10 GMT</pubDate>
            <atom:updated>2026-10-12T13:04:10.078Z</atom:updated>
            <content:encoded><![CDATA[<p>Hello avid picklers! I ran across an interesting situation that I, of course, wanted to share with you in case you need to do something weird like this too. Make sure the restraints are tight and fully locked, we’re going super-niche.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/511/1*r-79el_tNS10PRlb_-3pxQ.jpeg" /></figure><p>Let’s say you have something like this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Qp9rexFH9cDNdIsOGvazUg.png" /></figure><p>Yeah this is similar to the problem <a href="https://medium.com/@sunaj10/azure-virtual-network-gateways-the-great-connectivity-conundrum-cbd80446ff47">here </a>but this time we’re going to take a slightly different route. This time your boss tells you “Hey, we have this other vnet with a site to site gateway in it and we need to peer that to our hub vnet so they can get to onprem down our expressroute.” You reply “Sure, Chief” and proceed to have some inane banter about the weather for as long as possible to avoid any actual work.</p><p>Once you realize that that you can have the expressroute advertise the site to site range thanks to the “summarizedGatewayPrefixes” you think it’s fine to just peer the vnets and all will be well…until you do and realize that your Azure firewall doesn’t know how to get to the site to site range. Now you’re thinking “can I route directly to the gateway in the other vnet?” The answer, my friend, is *yes. You might be thinking of doing this via the “VirtualNetworkGateway” service tag, but that would just send traffic to the gateway in your vnet already. And this trick only works with VPN gateways, not Expressroute. You can route to the private ip address of the VPN gateway and treat it like an NVA, so for this situation you’d enable the gateway private ip then check the “connected devices” in the vnet and grab that ip. Then you can pop that into a UDR on the azfw subnet as the next hop for a route to the S2S range.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*L_b8SaqRPkk5gsUQDThJOw.png" /></figure><p>With that, the magic happens and traffic bounces off the gateway and on it’s merry way to the S2S tunnel. Leaving you and your boss in this situation:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/500/1*iZMalAdxT6Aq8-Vqqjriog.jpeg" /></figure><p>I hope this helps or at least shows you some of the weird things you can do with Azure routing. Thanks for reading through this and check out my Github for more labs and examples here: <a href="https://github.com/quiveringbacon">https://github.com/quiveringbacon</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/352/0*5c7DQQgBcnFW4qLK.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f7460d3742e6" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Azure Expressroute Troubleshooting: is it my problem?]]></title>
            <link>https://medium.com/@sunaj10/azure-expressroute-troubleshooting-is-it-my-problem-670081a710c3?source=rss-3e67c661019c------2</link>
            <guid isPermaLink="false">https://medium.com/p/670081a710c3</guid>
            <category><![CDATA[azure]]></category>
            <category><![CDATA[azure-networking]]></category>
            <dc:creator><![CDATA[Rusty Williams]]></dc:creator>
            <pubDate>Wed, 07 Oct 2026 23:13:44 GMT</pubDate>
            <atom:updated>2026-10-07T23:13:44.610Z</atom:updated>
            <content:encoded><![CDATA[<p>Ahoy mates! Let’s walk through a few scenarios where your expressroute private peering is borked and you need to find out what’s wrong while getting yelled at, aka just another Tuesday. Let’s check out some common issues and where to look first.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/360/1*-NKTrZcAiVFUfelNZDPJIw.gif" /></figure><p>Let’s start with the absolute basics, I talk about what an expressroute is and the peerings <a href="https://medium.com/aws-in-plain-english/the-need-for-speed-azure-expressroute-94a9b2db4b93">here</a>. Depending on your role and your company you might have access to both ends, meaning the onprem CPE and the Azure side, circuit and gateway, etc. But you might also only have access to the Azure side, so I’m going to focus on that end.</p><p>The first thing to check, is the circuit provisioned and enabled? You can glance at the portal and double check this.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/795/1*GBexD-JEMJEI5i7oj-55sA.png" /></figure><p>Are you getting ARP? For this you need to crack open either cloudshell or login to your azure account locally and run these powershell commands:</p><pre># Required Variables<br>$RG = &quot;&lt;Your Resource Group Name Here&gt;&quot;<br>$Name = &quot;&lt;Your ExpressRoute Circuit Name Here&gt;&quot;<br><br># ARP table for Azure private peering - Primary path<br>Get-AzExpressRouteCircuitARPTable -ResourceGroupName $RG -ExpressRouteCircuitName $Name -PeeringType AzurePrivatePeering -DevicePath Primary<br><br># ARP table for Azure private peering - Secondary path<br>Get-AzExpressRouteCircuitARPTable -ResourceGroupName $RG -ExpressRouteCircuitName $Name -PeeringType AzurePrivatePeering -DevicePath Secondary</pre><p>Depending on which type of MSEE you’re attached to you may only see the onprem mac address:</p><pre>Age InterfaceProperty IpAddress  MacAddress    <br>--- ----------------- ---------  ----------    <br> 10 On-Prem           10.10.10.1   ffff.eeee.dddd<br>  0 Microsoft         10.10.10.2   aaaa.bbbb.cccc</pre><p>If you don’t see that or you get an “Incomplete” for the onprem ARP then it’s your provider’s problem or something on the CPE. Double check that you’re using the first ip of the /30 onprem and the Ctag and Stag are correct. MS can only help you from the MSEE towards Azure, not from the MSEE to onprem. It’s up to you/your provider to get you to the MSEE.</p><p>If ARP is ok then move on to BGP and routes, you can either check the portal or do powershell but you’ll want to make sure the peering exists on the circuit and has the correct ip and ASN info and shared key if needed.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1020/1*FZH0oqmO40ulWYIIjEvODg.png" /></figure><p>If the peering is ok, then check routes on the circuit, also in the portal or powershell and see if you are getting routes from onprem and sending routes from Azure (under the peering section on the portal). If you are, is a route missing? If that’s the case then make sure that onprem isn’t advertising any overlapping ranges and make sure any peered vnets are using the hub vnet gateway in the peering settings.</p><p>If BGP is down but the peering settings are correct, make sure that onprem isn’t sending more prefixes than the limits allow and getting shutdown (limits <a href="https://learn.microsoft.com/en-us/troubleshoot/azure/expressroute/expressroute-troubleshooting-expressroute-overview#test-private-peering-connectivity">here</a>). If BGP is down but the limits aren’t hit, I still recommend checking with your provider first to make sure they don’t have something misconfigured but if they’re good then it’s time to open up a case with MS support.</p><p>If everything is really ok circuit-wise then it’s time to look at the gateway and vnet stuff (if you’re using private peering that is). Check the gateway and make sure it’s healthy. Take a look at the routes learned by the gateway and make sure they’re what’s expected. You can use these powershell commands to check if the gateway is peered to both MSEE’s and see the learned routes from them:</p><pre>Get-AzVirtualNetworkGatewayBGPPeerStatus -VirtualNetworkGatewayName gatewayName -ResourceGroupName resourceGroup<br>Get-AzVirtualNetworkGatewayLearnedRoute -ResourceGroupName resourceGroup -VirtualNetworkGatewayname gatewayName</pre><p>If the gateway isn’t peered to both MSEE’s or learning all the expected routes, then the gateway might be having a problem and you’ll need to either A. open a support case, or B. delete the gateway and connection and re-create them. Yeah that’s downtime but depending on how broken it is and if you care about an RCA, it’s by far the faster and easier thing to try.</p><p>If you’re using UDR’s to send traffic to an NVA or Azure Firewall, check the gateway subnet make sure the UDR is associated and that the right routes are in the UDR. You’ll need to check the other end and make sure the UDR on the spoke side also has the correct routes to avoid asymmetric craziness.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*aRz4HaVcu0RfhRtKIe2CBQ.png" /></figure><p>If you aren’t using UDR’s, take a look at the effective routes of VM nics in the portal or powershell and see what they are. Check if you see the onprem routes there (don’t worry if the next hop ip is a weird one, it should be the MSEE ip address that you won’t find anywhere). If you don’t see them then check the vnet peering (assuming the vm is not in the hub vnet) and make sure the peering settings are correct.</p><p>There you have it, a quick primer on Expressroute troubleshooting. Thanks for reading through it and I hope it helps. Check out my Github for more labs and examples here: <a href="https://github.com/quiveringbacon">https://github.com/quiveringbacon</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/352/0*B0SRvM9CxTcwPUCq.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=670081a710c3" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Azure Virtual Network Gateways: The Great Connectivity Conundrum]]></title>
            <link>https://medium.com/@sunaj10/azure-virtual-network-gateways-the-great-connectivity-conundrum-cbd80446ff47?source=rss-3e67c661019c------2</link>
            <guid isPermaLink="false">https://medium.com/p/cbd80446ff47</guid>
            <category><![CDATA[azure-networking]]></category>
            <category><![CDATA[azure]]></category>
            <dc:creator><![CDATA[Rusty Williams]]></dc:creator>
            <pubDate>Thu, 01 Oct 2026 12:28:48 GMT</pubDate>
            <atom:updated>2026-10-01T12:28:48.517Z</atom:updated>
            <content:encoded><![CDATA[<p>Greetings connectivity counselors! I wanted to talk to you today about a particularly weird problem that you might run into. No, not the 10 hotdogs but 8 buns issue, I’m talking about connecting hub vnets together or even a vnet to a vhub. It looks so innocent and simple even, but just like this little adorable bunny, it’ll F you up in a heartbeat. I do apologize as this one will be a bit long winded.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/470/1*nh0M6stJf6Cs8AGlrBHB2g.jpeg" /><figcaption>Just a harmless little bunny eh?</figcaption></figure><p>Imagine you’re presented with this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0Q_2GKkHUe_GbhnfHYe74w.png" /></figure><p>and of course you’re told “we’ll keep everything separate but we need to connect the 2 hubs together.” I know, you’re thinking “No problem, we’ll just peer the vnets” and yeah that’s all well and good but since both hubs have gateways you’ll literally just be peering the 2 hub vnets and onprem1 won’t learn about spoke2 or onprem2, maybe onprem doesn’t matter but they really need to get to spoke2 from onprem1. You actually have a few options now.</p><p>Yeah, the easiest way would be just get rid of that extra hub and connect the expressroutes and spokes to just 1 hub and be done. That kind of thinking will get you fired!</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/360/1*TBO1uxc2MzD9FmJzKieemA.gif" /></figure><p>There is SOME requirement that you keep both hubs so let’s ditch the easiest solution. I know, you’re thinking “but we can just connect the 2 expressroutes via global reach and be done”. Yeah that’s another easy way, but your boss is saying that traffic needs to go through BOTH firewalls, and nixes that idea. “On a scale of 1 to 10, we’re putting security at 11” they say. Can’t have you being spoiled by taking the easy way out.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*GK3c4icYFAEPklRCjJazMw.png" /></figure><p>That still leaves you a couple of other options, ranging from the completely ridiculous to the mildly crazy. The crux of the problem is really getting the routes to onprem. Normally you’ll peer spokes to hub vnets and use the hub (remote) gateway in the peering. This makes the gateway advertise the spoke ranges back to onprem. In this type of situation, you can’t use that option since both hubs have gateways, so the vnets will only know about each other and nothing else.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/360/1*EPlFu2L199njAAEWATFLQQ.gif" /></figure><p>The first stop on this crazy train is an ipsec tunnel between the vnets. No, you can’t just use the “vnet-to-vnet” option, that still won’t share spoke/branch info. It’ll have to be a big boy vpn with the whole gateway/local network gateway/connection trifecta on both ends. That doesn’t seem so bad, does it? Well if we want to transit from expressroute to vpn (and we do want that) then we’ll need a route server as well, and for that we need an active-active vpn gateway, AND it has to use 65515 for the ASN. So if we want both onprems to reach each other, well, then we’ll need that on both sides. Meaning NO BGP! (maniacal laugh) Yup, we’ll be using old fashioned static ranges in the local network gateway to have the VPN gateways know where the opposite side lives and have to update those if anything get added. You’ll still need UDR’s on both sides sending traffic to the respective firewalls. You could also use NVA’s instead of vpn gateways and get the same result but use bgp between them and route server to give routes to the spokes/other gateways.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*19GGV28LeSSKRqrrtbMFsA.png" /></figure><p>Moving to the next ring in this circus, I mentioned in another article <a href="https://medium.com/@sunaj10/azure-virtual-network-gateways-say-what-i-tell-you-to-d0f0ed281e94">here</a> that there’s a nifty new feature for vnets, the “summarizedGatewayPrefixes” property. With this little gem you can tell gateways what ranges they should advertise, doesn’t matter if they really know them or not. That means you can just peer the hubs without using the remote gateway setting but have each side advertise each other. Then you’ll still need a UDR-pocalypse of route tables…going from each gateway to the local firewall and another set for each firewall to get to each other for the remote ranges. Yup, just like the VPN gateway option, you’ll need to update each one if something changes.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*83ho9_pw0aNx3wgiNubuwQ.png" /></figure><p>Personally I actually prefer this UDR-palooza, yeah it’s a lot to touch but if the environment never changes then who cares? The updates are very fast to deploy at least and you don’t need to add more resources.</p><p>That brings us to the end of my little rant about connect vnets with gateways, thank you for reading through it and I hope it helps you. Check out my Github for more labs and examples here: <a href="https://github.com/quiveringbacon">https://github.com/quiveringbacon</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/352/0*B0SRvM9CxTcwPUCq.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=cbd80446ff47" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Azure Service Endpoints and Private Endpoints: straight to the (end)point]]></title>
            <link>https://medium.com/@sunaj10/azure-service-endpoints-and-private-endpoints-straight-to-the-end-point-2e404ade4e33?source=rss-3e67c661019c------2</link>
            <guid isPermaLink="false">https://medium.com/p/2e404ade4e33</guid>
            <category><![CDATA[azure-networking]]></category>
            <category><![CDATA[azure]]></category>
            <dc:creator><![CDATA[Rusty Williams]]></dc:creator>
            <pubDate>Thu, 24 Sep 2026 01:52:01 GMT</pubDate>
            <atom:updated>2026-09-24T01:52:01.067Z</atom:updated>
            <content:encoded><![CDATA[<p>Howdy everyone! Let’s talk about endpoints…specifically endpoints in Azure where you get not one, but TWO flavors: Service and Private. Both are useful but very different, let’s walk through some of the differences and why you would use one over the other.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*tZvxfdcrmJE3w1FrXIbPKQ.png" /><figcaption>funny AI generated image</figcaption></figure><p>Let’s deal with service endpoints first. Service endpoints let you put an endpoint in a subnet that lets you send traffic directly to the public IP of an Azure service (think storage, keyvault, etc..list of them <a href="https://learn.microsoft.com/en-us/azure/virtual-network/virtual-network-service-endpoints-overview">here</a>) <em>without </em>your VM having a public ip. Neat huh?</p><p>I want to address something right off the bat…you’ll see this all over the place, “Service endpoints keep your traffic on the Azure backbone and off the public internet.” Let’s think about this for just a few seconds, if my VM in Azure has a public IP (or some internet access) and I try to reach, let’s say a storage account via public access, do you think the traffic is leaving the VM and going out to some ISP and then returning back into Azure to reach the storage account?? I’m sorry to burst your bubble if you do think that, but, no that’s not what happens. Traffic still stays within Azure the whole time. So this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Vu1PUU_bwe5_j9-6izM88A.png" /></figure><p>versus this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*g64nMfoIpxKUXdfPE8J3wg.png" /></figure><p>The Azure service DNS entries still resolve to the public IP’s but now you’re able to restrict access on the service side to just whatever vnet has the service endpoint and not leave it open to everything on the internet, that’s the real benefit of service endpoints, the access restriction. Another thing to keep in mind is this is for vnets to access the Azure service, not onprem…you might be able to work around this if using Expressroute with MS peering, but really just plan on Private endpoints if you need to use them from onprem.</p><p>Now let’s have a look at the other side of the endpoint coin, private endpoint. A private endpoint gets a vnic in your vnet with a private ip. Privatelink attaches that private endpoint to a service, like a storage account. With that attached you can completely remove public access to the resource (like a storage account) and access it only via the private endpoint.</p><p>!!!Spoiler Alert!!! You’ll need to mess with DNS to get this working since the service hostname will still resolve to the public ip. You’ll want a private DNS zone setup with the private ip of the endpoint so it resolves to that rather than the public ip. More on that <a href="https://medium.com/aws-in-plain-english/azure-vwan-routing-15-private-painpoints-err-endpoints-ecb562bca0a1">here</a> and <a href="https://medium.com/aws-in-plain-english/azure-dns-privately-resolving-your-problems-55ce1e7ba380">here</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*x3ILY3ihYdVYuNWUAzAh5Q.png" /></figure><p>Bottom line, if you need to be completely “no public access” then you’ll need to go the Private endpoint route and have DNS setup to keep everything private and happy. Otherwise Service endpoints are easy and let you secure access quickly.</p><p>Thanks for reading through this and check out my Github for more labs and examples here: <a href="https://github.com/quiveringbacon">https://github.com/quiveringbacon</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/352/0*21ErVT5EvL3_LzNI.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2e404ade4e33" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Azure Load Balancer: Sharing the road]]></title>
            <link>https://medium.com/@sunaj10/azure-load-balancer-sharing-the-road-d5cad65ed05e?source=rss-3e67c661019c------2</link>
            <guid isPermaLink="false">https://medium.com/p/d5cad65ed05e</guid>
            <category><![CDATA[azure-networking]]></category>
            <category><![CDATA[azure]]></category>
            <dc:creator><![CDATA[Rusty Williams]]></dc:creator>
            <pubDate>Wed, 16 Sep 2026 23:56:12 GMT</pubDate>
            <atom:updated>2026-09-16T23:56:12.150Z</atom:updated>
            <content:encoded><![CDATA[<p>Heya! I noticed I haven’t shared anything dealing with Azure load balancer. Probably because I just don’t deal with it often, but let’s go through the ins and outs anyway, I’m sure you have some time to kill.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/360/1*L-6I-u1l4FFIK5RMPx5TbA.gif" /></figure><p>Azure load balancer pretty much does what you’d expect, traffic comes in to a frontend ip then gets sent to a vm in a backend pool. Give the frontend a public ip and it’s a public load balancer, give it a private ip and it’s an internal load balancer. You can’t mix the two and have a load balancer with both public and private frontends, but you can have multiple frontends.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*CqHl46Kpklrp8zvRVruPCA.png" /></figure><p>And the frontend (of an internal LB):</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ONRVVmDHOnVERwmFjjb_Jg.png" /></figure><p>Next you’ll need to create your backend pool of VM’s to send traffic to, you can either add the vNIC’s or the ip addresses of the VM’s (make sure any NSG’s allow this to the VM’s of course):</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*vLlloFQJWI7phTIU4sJanw.png" /></figure><p>And of course you’ll need to add a load balancing rule or an inbound NAT rule</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Yf2SCbnMuYf-Cq-ERjNnDA.png" /></figure><p>Basically, a load balancing rule is when you want traffic coming into a front end to go to a pool of VM’s, web servers for example. A NAT rule is if you want to port forward, traffic comes in to a frontend on a certain port then gets sent to a backend pool or specific VM on whatever port you specify.</p><p>You’ll also need to create a health probe:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*hCfwRpkhlzR8dzV5U0R1VA.png" /></figure><p>So that the LB will know what VM’s are available to receive traffic. These can be TCP, HTTP or HTTPS probes to specific ports on the VM’s.</p><p>Floating IP is something to note, in a nutshell what this does is change the destination IP that the backend VM gets. When it’s off (the default) the backend vm will get a packet with the client ip as the source and the VM ip as the destination. When it’s turned on the VM will get a packet with the client ip as the source and the LB frontend ip as the destination, you’ll need some OS configuration to allow this to work fyi (like a loopback with the frontend IP). If you’re using a public LB you can add outbound rules as well to NAT out to the internet (more info <a href="https://learn.microsoft.com/en-us/azure/load-balancer/load-balancer-outbound-connections">here</a>, NATgw is usually a better option).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/360/1*TZJzpRB1GzIVJLWv0LLr0g.gif" /><figcaption>exciting huh?</figcaption></figure><p>Session persistence can be set if you need that, by default the LB uses a 5-tuple hash. It uses the source ip and port, destination ip and port. You can change this to care only about the source ip or the source ip and protocol.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*iBeCzy5Yp5_yYQxCmVIkHA.png" /></figure><p>That’s really LB in a nutshell, I just wanted to run you through the setup and explain some of the options. Thanks for reading through this and check out my Github for more labs and examples here: <a href="https://github.com/quiveringbacon">https://github.com/quiveringbacon</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/352/0*21ErVT5EvL3_LzNI.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d5cad65ed05e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Azure Route Server: Route-mappin’ and Toe-tappin’]]></title>
            <link>https://medium.com/@sunaj10/azure-route-server-route-mappin-and-toe-tappin-bf90357a125a?source=rss-3e67c661019c------2</link>
            <guid isPermaLink="false">https://medium.com/p/bf90357a125a</guid>
            <category><![CDATA[azure]]></category>
            <category><![CDATA[azure-networking]]></category>
            <dc:creator><![CDATA[Rusty Williams]]></dc:creator>
            <pubDate>Thu, 10 Sep 2026 12:25:37 GMT</pubDate>
            <atom:updated>2026-09-10T12:25:37.744Z</atom:updated>
            <content:encoded><![CDATA[<p>Hi everybody! Let’s have a quick chat about Azure route server’s new feature, route maps! You may remember this particular feature from such products as Azure Vwan and my article on it <a href="https://medium.com/aws-in-plain-english/azure-vwan-routing-4-the-goblet-of-fire-578b723add86">here</a>. Now let’s talk about how you can use it on the less popular and successful brother to Vwan, route server.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/500/1*s2DDenxFmgGFH4LwpgNeYQ.jpeg" /><figcaption>Please understand, I actually prefer Luigi, he’s just not as popular with most folks</figcaption></figure><p>As part of Azure route server’s creation, you can imagine cutting the heart out of Vwan (the route service), stripping away the routing and connections and you’re left with route server. You’ll even see a lot of API functions for route server still refer to it as a vhub. You can BGP peer with it and it will plumb routes into vnets (vnics) and share routes with gateways…end of story, that’s what it does. Now route maps have trickled down from Vwan as well.</p><p>A major caveat here is that to apply route maps to any bgp, expressroute or vpn connection, your nva or gateway actually has to be in the same vnet as the route server like this:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0u0Vl_AQo_J0H0PJlkCisw.png" /></figure><p>The route map creation and conditions are all the same as in vwan, you can match against ip prefixes, ASN’s or bgp communities and alter or drop those as well. The route map itself is just a container for the route map rules that actually do things. Also like in vwan, the creation of the first one can take around 30 minutes or so, definitely plan for a long deployment time.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*VCfHbnTuYuPtYARV9EvFnw.png" /></figure><p>After creating them, you need to attach them to a connection, either inbound or outbound:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*7wtJv5Ls6wpMZghJ4IfVpQ.png" /></figure><p>Once that’s done, you’ll be rockin’ whatever changes you needed and I’m sure praise will follow.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/403/1*TDxiM-3dGf16BQByR93Mkg.jpeg" /></figure><p>Using route maps on route server can come in handy when needing to manipulate the routes coming from onprem via expressroute or site to site. For bgp connections to NVA’s, I’m not sure folks would use this rather than just use the NVA itself to do the route manipulation but hey, it’s great to have options. My colleague Jose Moreno talks about a particularly complicated setup using route maps on route server <a href="https://blog.cloudtrooper.net/2026/08/12/the-square-design-with-hub-and-spoke/">here</a>.</p><p>Thanks for reading through all this and check out my Github for more labs and examples here: <a href="https://github.com/quiveringbacon">https://github.com/quiveringbacon</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/352/0*NjYDGcx6sNuXcuGs.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=bf90357a125a" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>