all right thanks so I'm Donna schinsky CTO global sales organization we're going to talk a little bit more about SD when and the use case that we call us to win and of course I'm crazy enough to pull up another live demo that is out there running on live internet and lots of other uncontrollable network pads so we first of all I guess I wanted to know we've been spending a lot of time inside the box today if you think about it you know we've been talking about metadata and you know how we transfer that between nodes and and I guess you know what I like to do I guess with the SD win use case is kind of take a step outside the box right talk a little bit more about you know an end-to-end enterprise architecture what this would look like in the real world maybe talk about a little bit of the business value that we can bring to the SD way and use case and so you know this is this is really based on a tier well large enterprise fortune 500 enterprise architecture where they were looking to deploy potentially a combination of nodes in the branch sites retail branch sites they were looking for a good combination of price performance but also flexibility for the future so they literally couldn't decide between running a virtualized environment on the branch and then running an appliance version on the branch and interestingly enough you know we we can run of course any other form factor and so in this case you're seeing a VM host that it's literally paired H a with an appliance version right and so obviously typically a customer wouldn't choose both platforms but you know one in the sense of appliance reliability may be integrated LTE the other one a VM host with the ability to deploy applications dynamically out you know within the enterprise we've had another customer in fact kind of shrink down that VM host and and really it's an IOT use case where they're running docker on very compact appliance right so you can get these these appliances down to you know $300 and less in the in the branch office run containers right and then we can do the same form of segmentation for containers running in a docker host at the branch another interesting aspect of this use case is his service chaining I'm not sure we spend a lot of time talking about that today we did talk about it I think maybe in the context of Z scaler the ability to service change in third-party services we're not normally in the business of decrypting our customers traffic but you can embed a proxy in in that in that node through the service chaining so Palo Alto for example with a vm 100 is a partner and so that would be in an appliance form factor again those can run and KVM can be chained in using the tenet and service model we had a customer very interested in existing way an optimization gear they didn't want to throw it away it was a big investment we can service chain in third-party tools such as when opt but then that session context right so we think about the session context of secure vector routing can carry that way an optimization context up to a data center and sort of reversed service chain it right up at the data center so again the session awareness can be extremely useful to carry contacts from a branch office to a central office and in fact and sometimes I you know I say the flexibility there to run local security services centralized security services and do that on a per application basis is a huge win for the Sdn use case again you can have Wi-Fi and Wi-Fi context every single branch and every single branch office can have essentially a single tenant definition somebody's I think focuses on on wireless but that single tenant definition is global in nature so if I define a Wi-Fi tenant in terms of an access point or I define that in terms of an ingress interface if I want to make it change to the policy for a routing of Wi-Fi traffic that can be done globally and when we talk about being done globally and we talk about there may be services available and reachable in the data center right in this architecture I've got a central 120 80 nodes I also have data center based 120 80 nodes in this case I'm also showing again real-world multiple private data centers and in fact and this is kind of a combination of secure vector routing as well as traditional BGP and there was a little bit of discussion about overlay and and and an underlay this is essentially a blending of overlay and underlay we can run BGP over svr right and represent a data topology in the data center using BGP as some enterprises - right you may have existing nodes in a very large enterprise you may have back-end paths they may be using a s path prepending these are things that you can't come in and change necessarily overnight so in you know in a truly you know a router based sty architecture you can now you know introduce 120 80 I would say it's kind of the blending of the of the overlay and underlay so essentially you can do traditional routing when you don't have any other need or reason right and and secure vector routing when you want to do tight path control and quality of service and you know and QoS you know and and prioritization and Link failover right and so this blending I find so many Sdn solutions out there just you know they're just lowest common denominator tonal based load balancers right and and so having a true routing solution and then the ability to override traditional routing with secure vector routing is a huge win right and so when might we do that we might do that for cloud-based traffic right so we can have a 120 80 instance in the cloud that's literally part of the st1 i'll show you that in demo topology and and again that's treated like any other node in the network right so now we blended our enterprise right with the cloud that can be BG your bgp free I will show in fact later on we've got a cloud demo where we'll spin that up dynamically and then cloud gateways you find in some sd1 architectures it's optional it can be a path for steering traffic out an alternative egress interestingly enough a customer through a use case at us and they said well what if I have guest Wi-Fi you know many many sites and overloading my internet path right and you know how do I deal with that interestingly enough the tenant service model we think of it as being very granular and segmented you know down to ports and paths but we can be extremely broad as well right so let's take a default gateway running through a cloud gateway let's take five default gateways let's take 2000 branch offices and we can literally assign and divide traffic from a branch office to multiple internet routes literally on the fly right so if you think of that use case of its it's not necessarily segmentation down to the port and protocol right which we think of hyper segmentation it's like having five different named default routes and I'm literally assigning traffic to one or more default routes in in the in the network and so that's a it's really a huge win for traffic management now try to show a little bit of that through the demo as well other I guess components were showing here our LTE internet satellite MPLS we can run BGP with the underlay we can roll around BGP as an overlay so we integrate extremely well with existing MPLS I think it was already discussed that we can do secure vector routing or not so we can learn routes essentially from the quote unquote underlay and and blend very well and again low overhead we work well over LTE we've done all the work and I don't know if it was ever discussed but we've done all the work to do NAT traversal dynamically keep pinholes alive do outbound only you can literally have Nats and gnats in the path of an SDN connection which may be a cable modem broadband and we can punch outbound only to cloud and data center resources another interesting aspect of this is that it can be multi tier right and so we talked a lot about secure vector routing being a huge win when we try to you know try to scale up right and so in the sto and use case you know so many of the largest networks are hesitant to dive into st1 because they see it as a scaling issue and tunnels many thousands of tunnels and of course the n square problem or even just trying to do state management now the wonderful 128 t solution can introduce an intermediate layer in an SD LAN and and help scale up that way other questions on kind of outside the box sd1 architecture will go more into cloud in the next session all right so multipath routing I think we discussed this again previously but you know it is a big part of the SD Wang use case right you know and we call it the use case right the SD way and use case because essentially we're doing Enterprise routing and cloud routing and Sdn within the wind so Software Defined when we are ensuring optimal path selection by spur service based on quality right and LTE can be a backup path will have that adjacency established over LTE and MPLS of course can be a primary path and we can route between wired and wireless we can have of course any combination of underlay or transport network we can name the transport networks according to neighborhood I think it was discussed previously that we can then create sort of arbitrary topologies using say we define MPLS Network number one an MPLS network number two as two different neighborhoods we can have two different hubs and different spokes for two different MPLS neighborhoods so creating your topology of hub-and-spoke full mesh is extremely easy in the data model when we do session migration I think it was pointed out but maybe not fully pointed out so again so we've got it's configured by user based on needs of the service so a particular set of tennis can have a particular set of services maybe office 365 maybe voice and then of course the ingress branch will monitor path performance I think it was emphasized that we're passing metadata as part of the initial sessions setup right and then when we have a failover were resending that metadata along the alternative path to do session correlation all right so part of that session failover is the sending of metadata along the new path what that does is it again it carries the exact same context with the session it carries a session identifier again your same five tuple will be set up along the alternative path if you've set up a NAT and egress nat on the head end of that session will be correlated and you'll literally be flowing across the same path you can have and I'll show in the demo you can have several different policies for traffic you can have non revert of or non failover traffic you can have failover you can have reverted failover and your session migration is fully within your control I'll actually demonstrate that live using a voice call and Su and you know so we've talked week you know again it's it's a huge part of the business value that we bring Lane demonstrated that we're we're running 30% less bandwidth but let's also talk about entropy and how that applies specifically in satellite use cases low-bandwidth use cases we've seen you know multi t1 branch offices in the SD way and use case we've literally been pulled in by customers that have said hey we've got some satellite links out there satellites actually increasing in bandwidth and performance there's all kinds of cool you know Leo mio Gio alternatives out there for for satellite transport and and they literally pull us in they say you know this tunnel this to you and stuff is not working well over satellite you know they're having trouble you know with encrypted tunnels and so you know we've in live networks seen you know as much as you know 7x improvement in the you know in the in the bandwidth and download speeds in the sd1 use case so again you'll see this still there are all kinds of places that you'll end up still seeing satellite tactical environments and DoD of course oil platforms you'll have a combination of microwave and satellite potentially LTE links out there so you know this this use case is not going away right we think you know the throw bandwidth that the problem right and you know in saving 30 percent is not such an issue but absolutely in the you know in the low bandwidth link scenario we're extremely strong I think it was mentioned also that we do window size scaling local acknowledgments on a per session basis so for sessions that do Traverse high latency links in this case you know 6 to 700 milliseconds we can do the local acknowledgments that you would expect out of a why optimization solution so again that's kind of leveraging the TCP intelligence of the bit 120 80 solution overlapping IP addresses I think we decided that we hadn't really discussed this but you could see this in a managed service provider trying to provide services to multiple customers but we do see this in enterprise use cases certainly through consolidation very large networks where address migrations are challenging and so it can't be emphasized enough how secure vector routing can be used to essentially now assign words right so if you think about trying to do this and you know legacy technology I'm going to have you know an edge vrf and I'm gonna have route distinguishers an import and export from an MPLS core instance this is service provider MPLS right and it's extremely complex to try to bring that sort of you know level of inter working between multiple IP address spaces to the sto and use case and so you know by by assigning literally ten and a and ten at B we've already separate we've got maned we've maintained context right so 3.3 3.3 for ten a day now literally has a different word that it's being used when we when we do when we send the metadata to the data center right so we can literally keep context between overlapping address space you could deploy the exact same IP address to all your guest Wi-Fi access points or any no overlapping retail locations you want to keep very consistent addressing and that'll all be sorted out and with context at the head in it sound like there is a question maybe but no all right so again I think I'm oh yeah there's one more slide yeah so I'll show this a bit more in the cloud demo how we dynamically spin up capacity that's going to leverage you know some of the zero-touch deployment and you know zero-touch upgrade features and those will actually be shown in the in the demo video but it absolutely is part of the SD way in use case right you want to be able to punch these things out repeatedly repetitively with consistency right and so the conductor acts as a single repository for configuration data you can then template eyes and propagate the same configuration much of the data model again the tenant and the service data model is inherited by the router so a router configuration in a very large scale SD window rather than being a set of you know again routing tables and and tunnels you literally just have interfaces and adjacencies associated with the router right and so inherits and a new branch just simply inherits the the same top level objects that that every other node has and then when you change a policy to maybe increase the priority of voice traffic or decrease the priority of video traffic that will be inherited across the entire data model so you know zero touch features and zero touch deployment and repeated deployment is a huge part of the sd1 use case for large enterprises retail applications and certainly supported by the conductor model interestingly enough the biggest challenge in scaling up to thousands of sites you know what turned out to be you know a conductor management rather than tunnel management right and where any other solution is going to have an extremely complex you know network of tunnels you know we had to do a lot of work again just you know scaling and performance and good user experience for many many thousands of branch sites so again zero touch upgrade I think it may have been covered but there's a great inventory system on the conductor that will tell you about the availability of new software until it and in fact any anytime a nodes brought on there's some auditing done and it will determine the suitability of the the node for 120 potentially even download 120 80 if it's a brand new instance that doesn't have a software load and then you know and then bring it up to speed and bring it onto the network so here's the part that I have many things outside of my control but we will do a demo live demo of what I call my sandbox and so here we have let's just go to the top level - border are you able to see that or actually why is that not switching between displays that's right yes they are in control all right so here we are let me increase the font size here this is what I call my sandbox yeah I take this topology it's it's literally kind of grown it's to become essentially an operational network I use it from a lot of my own purposes it is my SD when deployment if you will it's got lots of routers I've got some in AWS and Azure this is again the top-level dashboard now you could see based on the the internet wins whether you know there would be an alarm in any one of these nodes I think we talked a little bit about different platforms but here's an example of regardless of the platform whether it's an appliance field whether it's a virtualized field whether it's a cloud feel it's the same look and feel you know in the 128 e dashboard so here's an HHA node that's that's two nodes that are represented as one right so you can literally manage a high availability router to physical platforms two sets of physical interfaces that'll work that'll manage is one note here's one sitting out in an oil platform and it's got some simulated satellite traffic when we kind of again we brush the mouse over any one of these nodes again all these adjacencies were generated by your preferred topology I want a full mesh on MPLS I want a hub-and-spoke on AWS or Azure and that you know that is represented in the chart here these are this is a live node running in in Azure I call my Azure gateway it's appearing with you know several nodes and then here's a live node running in AWS we talked a little bit about the the data model I guess I want to give you a little bit more of a real-world use case where I talk about you know just one of the interesting interesting use cases that I was presented so I again guess Wi-Fi a good use case because a lot of enterprises are kind of debate about the level of control that they want to put over guest Wi-Fi it's a nice global data model you know in my case I I've assigned it you know a few services multicast video internet is egressing through Azure so I'm using an azure based gateway can have a set of services chained in within the azure cloud it is attached to the wipe car but I have an interesting service here called DNS alright so the use case was hey I want to put some DNS security maybe some filtering for some sites into my into my you know in this case guest Wi-Fi network of course you can do it across the enterprise network well I don't know what my tenant might be configured to use right and it's so common to use the Google DNS server right so in my case here I had a bunch of clients that were out there and already deployed using a DNS server based on IP address think about DNS migrations you know if I want to migrate my DNS server from Prem to cloud or to another service I literally I'm gonna have to put in a DHCP address maybe into my into my DHCP server maybe that clients going to cache the DNS address the 128 T fabric can literally catch DNS right so I can take DNS traffic to eight eight eight eight or whatever my gateway was that was assigned snag port 53 TCP and UDP and then route that completely essentially exception route that to traditional IP so again we may have a default route going out you know to the Internet but I can snag DNS and send it to a DNS proxy if you're doing that so so let's let's go to the this idea of migrating DNS services right you set up that rule and you start working through all of those areas where you have the the old DNS server configured do you then have the capability of seeing how often that rule is being hit to find thanks for the leader there it is so when I define a service so again the beauty of this name data model and having the data model kind of be part of the routing is that one the enterprise in the sty use case or any other use case security use case define something that they cared about it suddenly now tracked on a session by session basis using that same word so I didn't have to do anything else now I have a stats bucket essentially 800 stats bucket on every single router that's classifying that session as DNS and DNS you know it could be port 53 but it might be a particular subset of DNS so the way we structure our Sdn topologies is typically they're typically around whatever the business cares about right and and that's extremely valuable and then you get this dashboard as literally as a result and these are different machines I have around the network and they're querying DNS sneakily I took my 8 8 8 8 I route it to a DNS proxy I do some logging recording and I send it to a DNS filtering website using an any cast address and that's it I mean it took literally about 10 minutes to set up a brand new anycast DNS service on this network it was a pretty cool use case so I like to point that one out we do it on time now we're doing alright but not that great so all right so what else do own a show you know again we talked a lot about failover doing it doing it in action I have a sip client actually that's running out - it's running over the LTE network so voice over LTE it's going to hit a gateway it actually hits my AWS central gateway it's then I use secure vector routing to kind of buy pass some outbound gnats and do some address and they're working into a private network and I've got a sip contact server and I can place a call in to my live demo Network this is handy because it's portable that's out and reachable on the internet actually let me let it ring for a second I think my ringer will come on but you can see my I've called myself and if anybody recognizes asterisk music on hold it's a asterisk music on hold but I've literally taken a call secure vector routed it right through the cloud back into a private premise network and that gave me a little bit of control over steering it across paths so I have a couple of paths in the network broadband and MPLS and now I have a VoIP call that's running again I can get kind of real-time analytics and look at the bandwidths that are being sent that's actually being sent twice over the link and I can look at my policy so I have a global policy that is defined for voice very specifically interestingly we have a service class called telephony right so that defines your queue behavior right so prioritization dscp percentage of bandwidth is going to be allocated to VoIP I've got a hunt and I've got a session resiliency type of a failover this is the part we get background music just cool I like this we're gonna have a failover policy a reversible failover policy you may or may not want to you know failover back and forth between LTE for example a packet duplication and packet retransmission I think we we mentioned can clean things up in the in certain environments packet duplication to be used for for tear correction I have a service level agreement that says I'm going to base it on latency Lawson jitter maximum allowable loss two percent latency jitter I have a set of vectors right now this is the secure vector routing notion I'm going to route this over broadband with IRD of five I'm gonna route this over MPLS the priority of ten and I'm not going to route my VoIP calls in this case over the internet right so extremely simple policy definition and and that's it's all it takes I have a wide area network simulator here I'm going to fire it up and I'm going to inject some packet loss apply boy hardly read that I'm going to apply some packet loss to Ethernet five you can see this is starting to drop packets on that particular link I can kind of refresh that show that that's live come back to my 1 2018 networking platform I also have a dashboard here it's really a dashboard that's very specific to VoIP right because my business cares about VoIP if I drop a call it could be revenue right so lost revenue could be lost money and now I'm Mike you know 122 fabric is detecting packet loss on this VoIP path and as you can see my broadband is no longer suitable for my voice traffic and I will go back to my MPLS link so that's done from a mobile phone look no hands that's live so anyway that's that's a good kind of real-world application of web traffic in the cool Astra music on hold is his plank and hang up the call and that traffic will go away so we'll talk about customers I think is the next one we're going to talk about youth basements and customers that are really deploying this in the field so yeah besides me a question I'll throw out now because I think it goes into the customer conversation when we start talking when we when we look at this as a solution what we see is we see really too major areas of of administration there's the network side and then there's the security side and a lot of organizations that's going to be two completely different teams how does how does that play out as far as allowing the security to manage those security policies only right and allowing the network to manage the network policies do you guys see issues with that and kind of how are things built in a way to create those demarcation points of control that's a good question well flexibility is key in my mind and security means different things to different folks right so for an enterprise that's worried about leaking Social Security numbers right security is different than a point of sale that's worried about segmentation security and controls so those different enterprises will have different security tools that may already be in place but the security team versus the network routing team right and so we can introduce the visibility segmentation we can introduce an easier mechanism to carry context to existing security tools right and in which case the security team is still owning and operating and/or they become part of the you know security management team for the 120 router what you're seeing out there yeah and usually it happens like in a deployment we usually go through both things first the routing team chooses us and then usually we have a review with the security team and show them you know how they can so also we have our back so basically it allows you to create users who will have access to specific tools within the conductor we also because we're ICS certified we have to do the crypto admin the security keys can be managed separately by a crypto admin and no other admin can see it so you have all those you have all those context available to it okay having said that the division of security team and networking team and we are clubbing it together there is still the two teams to go through here but we talk to both of them usually in a deployment and and discuss and like don't mention if they have security tools like Palo Alto or other things in in play in the network then we take it take the context from them and also service chain them or keep them in the path or keep them only for layer 7 level layer 7 functionality and so on what about Z scalar we have a deployment with that it's the same idea it's just a different you know now the tools are in the cloud and and you're you know fifteen billion URLs that you're trying to filter are no longer and the Prem and and you can manage those with through cloud tools so that sometimes you see that as part of an SDN migration they'll actually switch security tools to a cloud-based URI that's it yeah ok ok I'm gonna talk about customers before we go to the cloud the multi cloud and the cloud that most these guys are gonna show what I wanted to do is take a couple of different customer examples and show you we already talked about some but I want to go in a little bit detail what we did with them why they see a savings with us and who they are I sort of again tried to take a different sets of customers and especially obviously the ones which are publicly referenceable there are examples we have taken our CMC networks so who CMC networks they are the largest fan African and Middle East carrier they have Harlan ten pops globally their network is pretty pretty fast and pretty spread out across all of Africa and Middle East they're currently deployed with us in more than 40 countries they are also building their core network using us for step two use step so that they can do the cool functionality we get talked about you know going through multiple hops and so on one of the challenges they face like I said their end customers like World Bank United Nation embassies coca-cola and others operating in these regions they have strict destinies with them and they need to make sure that they meet those SSL is similar challenges they faced before they started using us poor reliability and performance like I said in general the business model is that you know there is some some issues with the customer and you have to pay some penalties it was I think it's like an accepted model of doing business and they face these huge challenges they actually tried in the field with many many of the existing or other Estevan competitors which are there and they found that the tunnel over had caused a lot of issues they have a largely o2 network which they already have that's going and capsulation and when they put another encapsulation on top of that they're hardly sending any data and they find that the end customers they actually use a third-party tool the end customers use a third-party tool for monitoring and they the monitor was that was the end to end what do you say the SLA getting and they monitored that and they base their their judgment on that of whether they have to pay penalties or not the traditional SD van had a lot of issues and that lead to leads to poor customer satisfaction obviously they have deployed us currently we're we deployed like I said in more than 40 countries in this is the phase 1 diagram where we are what they did is they have pretty much - one of the few who took us directly to a POC they actually took us to one of the sites which had problems and said let's see you know what you will do here and let's see how it works I have in the next what happened in the POC in the next slide but ideas basically proof your things work phase one we were deployed in 14 countries and they have won some enterprise customers they now have a ga version of the product they use silicone boxes which are four core Intel Atom processors and they put 120 D software and the ship resin appliance to their end customers to deploy currently it's deployed in more than 40 countries and like I said C Ron is there their version of SD van they call it and they are working towards towards getting ready for step and also one thing they noticed is this now since they are able to offer good SLS to their customers they actually want to move away from doing capacity based charging or bandwidth base charging to actually charging based on the SLS they deliver they can flip the thing around and say you all originally is to tell us that we are not meeting the SL is what Howard we sell to you based on the SL is in the sense we will give you 10 Meg and we guarantee that or whatever in this kind of rates and this kind of drop rates this kind of latency and so on so they actually want to flip and base their services on the SLA is that way they can monetize better the services they are offering these are the sites they chose for the they chose us remote site in in in near Johannesburg and Swaziland and they put our outers in and they tested it one of the things we noticed is that our routers in case of degradation or latency if you look at it the first one shows bandwidth over Man 1 bandwidth over band 2 and latency as you look at it as latency changes we switch over the traffic to the other bandwidth link and we are able to deliver the SLS they require Lin was actually involved in this these actually his graphs he was involved in this deployment then he tried and tested it with them any comments on the on the user expedience they also had the feature request and we made it happen yeah well one of the things and we talked about the fabric fragmentation numerous times today but it was very very important to them based on customer requirements that they be able to deliver their service with a full 1500 byte MTU to the end customer so you know and we're encrypting it so when you do that it is gonna exceed 1,500 m to use of that fabric fragmentation is a very very important feature for these guys and they're they're pretty thrilled with it oh we did the same and this is another thing I took for the same thing for packet loss as you look at it as packet loss changes packet loss increases that's the same time we switch it over the back oh man - and we are able to deliver the SLS they want the end result for them they have improved improved throughputs insights which they have used to have poor performance and of course improved SLS overall benefits this improved performance zero downtime with failures because of the switchover better utilization of low-cost links and cost savings we have a case study written with them jointly written with them and it's up on our website with CMC and a managed service for partner we have called red wine actually there are links on these presentations if all you can look at our website but this is available on our website to read the other use case I wanted to talk about is converge one who they are they provide unified communication services to 11,000 enterprise customers not only unified communications there are many different kinds of services they provide Unified Communications is one of them are they have 11,000 enterprise customers this service 73 + 73 % which is Saturday 3 of the 1400 they have 2,000 employees they are big in the United States they are a leading provider of it like I said unified communications and other technology solutions one of their needs was what currently have that their original design was if you want you to fight communication services you have sites you bring your MPLS link into art into their data centers and and integrate with their voice back end and the main main concern they had with this kind of solution was that it takes a long time obviously for an MPLS connection to come into that datacenter and for the customer to bring their MPLS router in it also doesn't provide great SLA s at the end because the customer has to do a lot of the legwork to bring the C router in place the C router and here in the datacenter it's not really a managed service offering and the customer has to do all that work and of course they are stuck with using MPLS and high cost their circuits only they wanted to switch over to using internet paths obviously now they have using the 128 D solution they call it the the c1 SV router of converge one SD band router basically the please place the customers the 128 e routers and the customer sites and of course they have the 120 router in their data centers now they can provide connectivity over MPLS and Internet they also have the ability to do that over to internet paths if you have two ISPs with two internet connections they're able to provide and guarantee this kind of SLA is theirs they call this for small medium redundancy and for MPLS redundancy they have both classes of service they are offering to their customers currently or they have a number of customers they have one through this already they use the laner devices laner for core again the four core atom processors with our software on it for the deployment one of the things they like about the feature is that you're able to select paths based on mas course and they have that set in their network currently so what we do is we basically monitor mas mas scores for different paths in the network based on the delay jitter and latency that we see in different paths we calculate perceived mots ma schools for voice traffic if at any point we see that most amosmoss score declines we can switch over traffic to another path and the network the advantage is you can set more score you can just say keep it the over path which has 4 ma score of 4 at least all the time and it will do that one of the use cases they had it's also documented in the in the webinar I did with them they had they have an end customer which is a credit union in the in the US and they had launched a call center when they went live with convo one and the call center was having 800 calls the mpls provider unfortunately dropped their MPLS connection and so everything switched over to LTE and they didn't see a glitch obviously the MPLS provider looked really bad but we look really we came out as heroes it was a it was a good use case to see in the field that had happened and we actually switched over all the 800 calls without any without any issues I can watch one and we looked very good in the customers eyes in front of the credit union but this is a life case in which it happened in the network and they actually routed all the calls over some of the benefits they see and the benefits they offer to their customer SLA extension now they are able to provide SLA is over Internet to any any customers because the software based solution they see advantage of that like I said they're using the Lanner units they order the liner units through a partner of ours called arrow and they use it to deployments this session aware they see all the benefits of fast fail overs obviously the use case where the traffic switched over was was one of the big things and of course since they're providing enterprise communications they need to guarantee a zero trust network so it helped them to have Phipps compliance and and PCI compliance and other things in their network the other use case I wanted to talk about is bard materials they are a construction company in in the u.s. they're they have mining locations in these three states they have 20 ready mix plants and 30 aggregate sources they basically get all these materials from different paths to their to their main locations one of the challenges they face is that the network is not very reliable and also it's they have very varied connections because these are some of some of these sites are remote so they are very varied then it's not like they have an isp who will provide a guaranteed link or some sort of SLA the current network they had the original network they had they would have sites they usually use the point-to-point wireless connection for their deployment and they also have DSL one and DSL two from some ins on some sites from their from their providers one of the challenges they face is that they always want to get a feed of the like a set of video feed which tells them how many trucks are passing the gates the reason they need that is the number of trucks tell them how much how much materials are coming into their sites if they don't have that feed they are in the blind they actually don't know what's going on and how much materials are going to come in today and what's gonna happen they can't really plan for it so they need that video feed at any cost so they want to what they want to do is they want to get that video feed they won't have instantaneous failover and then there's a failure they can't tolerate it they need to get that link to switch over and the video feed to reach there the obviously one morning do have that video feed have the highest priority in case anything happens in the network congestion they want to get that video feed they want a balanced load they wanna monitor monitor everything centrally and they want to limit I gotta get bandwidth and of course they wanna reduce costs and other things those those are common to everybody what they have done now is they have switched to using the 128 t solution they actually place our conductor in the AWS cloud they use QoS and T for rate limiting and priority zipper prioritizing that video feed and of course they have LTE for multiple connections and for remote sites they use LTE because of that they have instantaneous fail overs they are able to use the they also use Lana Lana boxes and the Intel four core processors on different sites and they have of course real-time monitoring and load balancing one of the advantages they see is obviously they are able to see that video feed all the time they believe that because of that they see a saving of 1k per truck per day because of that improved connectivity they get and they're able to monitor the feed and make better judgment about their business they of course have multiplied optimizations and now lower costs overall the thing they're saving about fifty percent compared to any other Estevan offering they had and finally the last example I want to give before I take some questions or switch over to the the cloud we switch over to the cloud demo is revision systems our relation provides unified communications to 300 hospitals in the US Canada and Mexico they have spread out over 20 states in the US and three countries since they are providing unified communications to hospitals they need to have 100% uptime or guaranteed SL A's one of the challenges they face obviously is that they don't control the network in the middle usually what happens is the hospitals have they call them up and say when there is a degradation of service oh Vanessa shoes with the calls they'll call them up in and and dr. evasion revision does debugging and so on finds out what's happening oh the ISP in the middle is not offering the service or you know has to drop the connection and stuff like that they usually unable to not they don't control the network in the middle so that leads to issues what they want is to extend their there their ability to deliver these services to the customer premise self this is this is what that deployment looks like they currently have it over multiple ISPs connected to different hospitals and to their data centers their data centers of course also hosts our conductor which tells them who's who's where they are able to get end-to-end control hyper segmentation all the other benefits we talk about one of the the the best use cases they have is that they say that now they are able to extend the SLA from their data center to the customer site we are able to guarantee those SLA is over a third party link again we have a a case study written with them and it's also available on our website you can read more about it how do you mean another Sdn solutions failed in what way most Sdn solutions like we we noticed at at least that in the trial which revision did is that whenever there's a failure in the network and it switches over and creates another tunnel and fails fails over usually the voice call drops and they have to reconnect the call it leads to poor experience for the customer in our case the failure was always instantaneous that is why they thought that you know it's more superior to their exist to the other has different solutions they tried in this deployment in most cases we have noticed like I said the two issues with the backup tunnels one is it either locks the IP address and you have to pay for data all the time from the LTE network you can create a backup tunnel and keep that up just send a ping once in a while and keep it up but then you have to do that and the other is basically when you do it on the fly it takes a long time it usually takes time to get the right the address from the LTE network and failover it usually works well when there's two wired connections but usually there's an issue in the LD network that's what we all noticed here
