 good evening everyone so we are going to talk about network visibility in hybrid and multi clouds so let's have a quick introduction about ourselves so that is my github ID I'm Suresh and I have around 15 years of experience in cloud telcom and geospatial technologies and I'm a cloud cloud stack contributors in last four plus years and currently I'm a software engineer at Axel right so we work on a product called cloud platform which is open source version cloud stack under Apache so this is what you can expect from our talk so we will go through the introduction of what network visibility is and just cover the challenges around the network across the clouds and we will cover few use cases and the business impacts out of that so these all based on our experience in the current product and federating between Apache cloud stack and AWS okay so so coming to interest so network visibility it provides a platform for like dig into the network and get to know the details in a granular level so it answers several questions like what are the resources in use and who is using them and you can trace the actual network flows using the network insights provided from the visibility and within cloud and environment so if a particular virtual machine has to reach another virtual machine in another cloud in hybrid or multi-cloud environment so you can get the details of that and you can check the various policies set and what are the policies stopping your traffic flow and what are the misconfigurations within our network rules so that you can reconfigure that and even you base from the network insights you can get the different traffic patterns and you can predict the future traffic and you can get to know like what is the next traffic after 36 hours or in a day so even with knowing the network insights you can get to know what are the tools required in the network so as to improve your network stability and so that you can integrate with other monitoring tools or any out-of-band tools so so basically at a high high level the visibility provides end-to-end traffic flow from one cloud to another cloud across high hybrid or multi-clouds with which you can know the actual data flow between the clouds which can be used for various analytics okay and in the in the hybrid and multi-cloud environment so there's a various network activities so this network visibility will give you a detailed report of the network activity within the clouds which administrators can take a look and use it for route causing or troubleshooting or improving their policies and maybe doing the better configuration to improve the speed of a particular network links and even you can get a like a broader review of the network patterns or the network flows within hybrid and multi-cloud environment okay so these are the challenges in the current hybrid and multi-cloud environment so basically this environment whatever within the clouds is a complex it has a multi vendor appliances and multiple service providers and it uses a diverse hardware and software platforms so it is hard to monitor each and every appliance so this visibility will provide a easy way so as to monitor each and every appliance within within the environment and you can get good insights out of that and network choking so if anyone wants to like the administrator want to know where actually the choke has happened so it is very difficult to figure out in a complex environment so and network patterns so if you want to check different network patterns from various clouds and what are the resources which is using a huge bandwidth in that network so we find finding with the current monitoring tools is quite a bit difficult with that and tracking the history so if you want to know what has happened last week within my cloud environment or what has happened in last day in my cloud environment so how do you track the history but you don't have anything to track the history so and suppose there was a security threat or an attack within your network so how do you track that so that is also one challenge challenging area in a multi-cloud and high-build environment and performance so as the there's a huge data volume and that network traffic flows between various virtual machines and the network appliances in the cloud and environment so the performance is also an important aspect performance and the scalability of the network so these are the challenges which we came across and with the network visibility the solution which my friend will propose so we can get some information about various insights into the network and just overcome these challenges so my friend will just go through the other details about the cloud environment thanks so before going into the complete details of network visibility so before going to the network visibility inside I just want to give you a quick intro for how cloud environment I mean what are all the cloud environments we have we have I mean on premise cloud technologies which means it's organizations organization have its own infrastructure they can build their own cloud they can manage themselves the other one is the public clouds like AWS or Google cloud platform or Microsoft issue they have their own infrastructure and they provide us some services so that we can use them so both have their own advantages and disadvantages so the mix of these two can be the hybrid clouds where most of the people are tend to use that like it's a for on premise cloud technologies we have the complete control of the infrastructure it's a something you want to have and our own deployment like where we want to keep the router switches and how the complete lab is under control and even we can deploy some network tabs or network aggregators within our infrastructure so even public clouds also they are easy and fast to spin up any applications say if I own a website if I develop a website on a public cloud I deploy a website if I want to I mean double that load I can easily spawn some other applications and make it work on the public clouds easily and same if we can do with on premise clouds also but it will be little easier and faster on public clouds so to mix both these clouds we can have the hybrid clouds so like I say an organization has a huge demand of traffic in at least I mean in any other if they have only one month having huge demand of the traffic they don't need to have on-premise infrastructure all over the year they can use this public cloud for that month itself so there the hybrid cloud comes into picture so coming to the network visibility across this hybrid cloud so before going into the use cases like I mean based on the challenges my friend has talked about so this is all based upon our experience on Apache cloud stack if say we have we get most of the escalations around networking area only I mean there might be some other issues or there might be some networking issues every time we have to go into the router whether it could be a virtual router or a physical router we have to log into the router and troubleshoot it individually that is the pain point so what we thought of we have looked for some this kind of network visibility applications there are many applications in the market but what we thought is to reduce the cost by using the existing flow traffic protocols there are a lot many flow traffic protocols which capture the network flow we can capture them and use within our applications so like if you have the network visibility upon our cloud we can as we said we can troubleshoot it properly and if say there are any blind spots say suppose if I have 10 routers in my network infrastructure and I'm not sure what happened to one of the routers the complete network has shut down or a VPC tier has not getting the traffic so in that case we can find the blind spots easily and even the policy complaints also most of the organizations are now I mean using those policy compliances so based on the data we get I mean because this network we capture that whatever the data we capture that so we can assure this policy compliances and placement strategies say say we deploy a hybrid cloud so we have both on-premise VMs and the VMs on the public clouds as well so based on the monitoring or analysis we done on the network traffic we can decide if a VM can be placed in on-premise VM or a public VM it might not be suitable on a normal on-premise infrastructure so we can move it to the public cloud same with the configuration as well we say the deployment that we have might need some configuration changes so with this network visibility we can change the configuration so that it works much better so the business impact is similar I mean I mean we have this network visibility we can enhance the customer experience and we can build the robust platforms by like changing the placement strategies or changing the configurations and even by finding out the network patterns we can figure out some of the security we can solve the some of the security concerns as well if there is any difference in the network pattern I mean usual network pattern so we can find out what happened and we can quickly go through and check it so I have not proposing a complete solution but this is the basic network visibility in a normal network infrastructure I mean there are a lot other network visibility applications but those might be costlier but as I told you there are many flow traffic protocols that we can use and make it working network visibility if say if say these are the rack of the service I have on which there might be cloud or the way can use it as it is and the network infrastructure on top of the service so this is the system I have so on top of it I want to have this network visibility around the traffic that is being flowing across these routers and switches so what we so the basic fundamental of network visibility is to capture the network traffic there are multiple solutions to capture that we can use either of the ways but I'm proper I'm just explaining some of the hardware and software things that we can use so say suppose the traffic from the service going through the infrastructure from the routers to the switches and the traffic flows between these servers I mean these can be VMs as well so what we can do is we can keep an end tap so this is a network test access point so this is an hardware appliance so this will simply capture the network flowing from the routers to the switches within the network infrastructure so the problem with these n-taps is that this is an hardware appliance and this might be costlier at times I mean based on the cloud deployment we have and there is another port port mirroring that we can use in the switches so what n-tap actually does is it will simply duplicates the network traffic without touching the actual traffic so it will just duplicates the traffic and sends it to the network traffic collector so this we have to implement n-tap only provides a way to send it to some endpoint so this and network traffic we have to write that and collect it and send it to the corresponding network monitor whether it could be troubleshooting or any security monitor so there is other way to send the traffic to the network traffic collector which is called span which is switched port analyzer so which is basically port mapping so as I said network n-tap does the complete network duplication but what span does is we can configure it per port I mean we can duplicate the traffic from port to port port A to port B so we can decide what are all the ports can be monitored and send that traffic to the network collector and from there we can forward to the monitors so if say I'm I'm totally talking about with respect to the network infrastructure it say I don't have the access to the network infrastructure it say I have a cloud deployed on these servers I mean as I said we worked on Apache cloud stack right this is an IAS platform infrastructure as a service platform so which provides a cloud on top of the existing servers so in that there are many virtual routers VMs and many other network appliances software network appliances so there I can't deploy these n-taps maybe I can configure a span but not n-taps so there what we have to do is in VMs there is something called virtual n-taps so we can I mean that's that's making use of the OS level capabilities to capture the network at the interface level and send it to the network traffic collector so instead of using the hardware appliances we can use the virtual n-taps so that's on the normal cloud if say if we have hybrid clouds so they say this is my on-premise cloud and this is my public cloud so what all I can do is so whatever the traffic flows with flows within this network I can capture it and send it to the monitors but the problem comes with the public public cloud I only have the access to the services or the VMs on the public cloud not directly to the servers so I can't deploy n-taps or even I can't deploy as span on the switches so what we can do is so even if say if we if we put a load balancer between this cloud private cloud and this public cloud I can only monitor the complete traffic grow going to this public cloud not each and every VM so the same thing that we were talking about the VMs deployed on the servers so we have to also have a tapping agent install on each and every VM we deployed on public clouds so what that will do is tapping agents will send the traffic to the network traffic collector and from there it will send it to the monitor so these tapping agents can be either simple Java agents or it can be any Docker containers or anything that is up to our deployment development so we need to have these tapping agents for sure on this public cloud so how we can achieve this is so since we in our Apache cloud stack we have done an integration with AWS so we have done a project called cloud federation which will actually unifies the cloud on premise cloud and the public cloud so we will have a unified so we don't even have two interfaces we'll have only one interface we can deploy the VM we can tell where it can be deployed we can directly deploy it on the on premise machine or on the public cloud so there we got this problem of network network visibility that's where we thought of looking into some solution and we are still working on it I've not yet completed but we are trying to deploy these kind of agents on the VMs so there in the tapping agents what we are using is there are multiple network flow protocols like net flow which is a Cisco proprietary and there are many other vendor specific network protocols so there we in that network protocol we are using that to send the traffic to the network traffic controller so the biggest question is this is an on-premise network and this is a public network so what we have done to send the traffic from the servers to the network traffic collector is to have a VPN VPN connection between these two servers I mean this again this collector will be deployed somewhere on the servers only so there will be a VPN connection so that the complete traffic will be duplicated and send it to the network traffic collector so this network traffic collector either this can be so whatever the tapping agents we have deployed here there should be a receiver over here also so in some cases we have deployed that receiver outside of the network traffic collector and we send the traffic and forward the traffic again to the network traffic collector and from then we'll send it to them network monitors so again the network monitors solutions there are multiple solutions based on the network that we capture we are we are using some other solutions not we are writing these solutions we are just trying to make this working on our own like deploying the tapping agents and sending it to the network traffic collector so we already have some span related development on our switches so that is already done but this this need to be done yeah so that's what we are working on maybe next sessions will be demo on that not now maybe in the next conference we are done if you have any questions