[0:00]So before we get to the next session, you know, one point I didn't want to bring up. So something that we've seen across all three days, as much as, I have a bias for NSO and we love NSO's a tool.
[0:12]I think the reality is automation is a team sport. It's going to be a combination of different tools that address different different user communities and different requirements. And, you know, one part of automation, you know, at scale, really is a systems integration exercise.
[0:27]You look at, you know, look at the customer sessions that we've had over the last three days. All of them have picked different tools that meet their requirements and they've been very thoughtful in saying, we're not going to pick, we're not going to use every single tool we can find.
[0:40]But, you know, they've picked a handful of tools and then, you know, spent some time figuring out how to integrate them and how to build workflows. And I think that's one of the secrets we see with, with customers that are successful with their strategies. And, and you know, something to think about, you know, on the on the ride home.
[0:55]So we're going to pivot a little bit. We've we're down in the weeds with test strategy and those kind of things. We're going to go work up up stack for the next session and talk about operations and and, uh, and even if you're a developer, uh, why you should care.
[1:09]So with that, I will hand things over to Moty. And take us away. Yeah, thank you, Omar.
[1:20]Let's see if the slides are running.
[1:26]Okay, hello everybody. Um, I think it's a privilege to be here on the last day because, uh, I got many questions, many feedbacks on, uh, things that I'm going to talk about.
[1:39]So, um, I'm Moti, I'm the product manager of the crosswork HCO, um, which is, uh, acquired, which is part of the Sedona acquisition by Cisco two years ago. Now,
[1:50]it's part of crosswork now. And, um, I think two years ago that many people didn't know what HCO is about. It was called Net Fusion back then.
[2:00]But now, I think most of you have heard something about the HCO. But still many questions came, like, um, how does it coexist with NSO? Because NSO is a, is a, is an automation of provisioning.
[2:12]Then in HCO, we also have provisioning, but, um, what about, uh, what about that part? How does it coexist? Um, it's part of crosswork, but I have CCNP. So why do I need HCO?
[2:24]And, uh, and what is the real value of the HCO? So, these are all questions that, um, being asked, and, uh, and I would like to take the chance here to, uh, try to answer them.
[2:35]So, um, no, that's not the right slide. So, let's switch here.
[2:51]Okay, then we'll use these ones. In any case, we started with, uh, Sedona and, uh, trying to actually answer, um, a big question.
[2:58]We get a network and a network is, uh, is a, is a full of silos. There are the IP aggregation, metro optical, and, uh, and other parts like microwave, for example.
[3:09]How do we connect all pieces together? How do we make sure that, uh, customers can see the full network? Um, getting the multi-layer, multi-domain, all information in, uh, in one system, that can be then used for, uh, um, for operations, for automation of, uh,
[3:22]everything, uh, everything around the network. And that comes to, uh, um, to a solution that we have developed with algorithms, to get all this information into one system.
[3:40]Which means that we have developed algorithms to take the information from the systems, from the IP network, from the optical network, and, uh, correlate all the layers together.
[3:50]And creating a full picture of the network from the services in the IP down to the IP topology, the optical topology, and even to the fibers.
[4:01]And we're able to show a full picture of the network from the service layer to the fiber. Now, that came into, uh, a product that then we called Net Fusion, and it's now the crosswork HCO.
[4:14]And, uh, the crosswork HCO has, uh, several advantages. First of all, it shows a multi-layer view of the network. A 3D view of all the network.
[4:24]And then, um, it has also all the inventory information. All the inventory pieces, like cards, pluggables, devices, links, they're all managed in one system.
[4:33]Um, we get many assurance capabilities. Um, getting performance of the network, getting the operational status of the services and the links and being able to show that, visualize that, and have interfaces to get this, uh, get this into other systems.
[4:47]We have analytics tools that are part of the system to mainly run what-if simulations for failure impacts, for, uh, traffic growth, for optimization,
[4:57]shared risk analysis because now that we see all the layers of, uh, the network, we are able actually to tell you what will happen if you disconnect a fiber, which services will be impacted, or, for example, if you have some maintenance, um, operations that you would like to, to run, you would like to know what will be the impact of a maintenance.
[5:16]And provisioning, and we do support provisioning. Then I'll come to provisioning and explain what exactly we support in provisioning, because that somewhat coexist with what NSO has, but it has differences.
[5:30]First of all, I would like to answer a question of, uh, where we are in the puzzle of the automation and the architecture that is, uh, being driven by all the standard bodies, like, uh, TIP, for example, ONF, ITF.
[5:45]So, as you can see in the, uh, architecture, um, there are controllers. There are the IP controllers, which, uh, we Cisco have the crosswork CNC. Um, there are IP controllers from Nokia, Juniper, Huawei as well.
[5:59]There are the optical controllers by the lead vendors, like Ciena, Nokia, Huawei, Cisco, and, uh, we're start to see also microwave controllers.
[6:07]And those controllers, they're managing the network and taking care of everything about the operations, automation, but then on top of them, there is the crosswork hierarchical controller.
[6:17]Which means that we collect everything from those controllers, we put them in one system, and we're able to provide you a full view of, uh, of, uh, the network.
[6:30]And, um, with that, what are the typical use cases that, uh, we see in, uh, in HCO? Because, um, when we talk about use cases, then,
[6:40]we need to understand actually what is, uh, what is the value? Because we do have CCNP. We do have, uh, optical controllers. And then, what are the cases when you actually need HCO to, uh, uh, to be there in the network?
[6:56]So, first of all, it's, uh, it's for analytics. So, as I said, we do collect data from multiple sources and being able to turn that into intelligence.
[7:05]So intelligence can be a lot of use cases to, uh, run vulnerability check, to, uh, to talk about the shared risks, um, what are the unused equipments in a network,
[7:14]and, uh, if you want to run optimizations, or to, uh, simulate failure impacts, then we do all that in, uh, in HCO. And when it comes on top of, uh, the controllers, it means that now you can run those operations not only in the single domain, but in a multiple domain.
[7:34]So, for example, in cases you have only, um, IP network managed by a cynical controller, and you are not connecting that, and you don't have an optical network, a photonic optical network, you don't actually need HCO.
[7:47]You need HCO when you have multiple IP, IP domains and when you have multiple optical domains, and you would like to connect them all together and see the full picture of, uh, of, uh, the network, then you need it. And it runs analytics, but in addition to analytics, it runs also assurance.
[8:04]It means that thanks to the fact that we have all the multi-layers, we're able to show your root cause analysis, meaning that when something happened to a lower layer links, we're able to tell you which services were impacted by the, uh, by the failure.
[8:21]And then to, uh, help you, um, accelerate the time that it takes you from understanding there is a failure to the understanding of which services are impacted, which customers needed to be notified and how to repair, to repair the problem.
[8:36]And in addition to that, we also collect, uh, performance, uh, metrics. So, the unique value that we have in performance is actually to take all the performance information,
[8:47]which is, um, information we collect from the controllers and from other parts of the network, connect that and correlate that with the topology.
[8:59]So, for example, if you look at the, at the LAG bundle, and you would like to know whether all the LAG members, they're equally share the bandwidth, actually, actually utilizing the bandwidth.
[9:08]You would like to see if there are some imbalances between those links. So, uh, we're able to correlate the information of the performance with the information of, uh, topology.
[9:20]And if you would like to see, for example, all the LSPs running from site A to site B, and to see the utilization of all these LSPs in a single report.
[9:34]So, this correlation between topology to performance is one of a, one of the key values, and, uh, we have an active inventory, and many customers do use active inventory in order to, uh, feed that information into their OSS, and into their automation workflows.
[9:50]It means that we get this information like, as I said before, cards, pluggables, links, ports, all in the same system.
[9:57]It's one, once you have it in the same system, then you can feed workflows like, for example, before you would like to run some service provisioning. You would like to know if the port is there, what is the status of the port from an operational perspective?
[10:08]If there is a need to order a new card, all this information is, is, is actually there. And the way to consume the information is very easy.
[10:19]And I will talk about, uh, a tool that we have, which is part of the HCO, called SHQL.
[10:22]It's a, it's a tool to allow you to consume this information in a REST API, very convenient to use and very innovative in a way that it runs over, um, several, uh, several, uh, um,
[10:35]dimensions in a network and get this information easily. So, um, in addition to, the capabilities of collecting information, having that in, uh, in a unified model, we have applications on top of the HCO.
[10:54]And the applications are for different purposes. First of all, we have a 3D map of the network.
[11:01]So, we're able to show you a network, which is going over the layers, and once you click on a link in an IP layer, you're able to see how the path goes over the optical network, and even also on the fiber path.
[11:15]Because once we collect the information on the fiber path, we're able to show you exactly if something happened in, in a fiber, then what are the layers, like the layer one layers and the IP layers that were impacted and using those links.
[11:30]And other applications are, for example, to get history of the model because that model that is part of, uh, of, uh, the HCO, um, is a model that is being updated all the time.
[12:19]Cards are being added, links failed, link added. And we keep snapshots of all this information throughout the time. So, once you have this information, and then you would like now to begin to history and understand if something happened, then you would like to, uh, use our time machine, which helps you to get back in time.
[12:40]Select a point in time in which you ask yourself, what happened in this point in time? How the network looked at this point? And this, able, this actually gives you the ability to analyze, um, issues that you have in the network, or, for example, to understand the consequences of events that happened throughout a specific period of time.
[13:03]And, uh, I would like to talk about the value of, uh, of the SHQL. SHQL is actually, um, the Sedona Hierarchical Query Language. And the problem that it aims to solve is when you have a model, which is, um, with a many dimensions.
[13:22]We have the time dimension, we have the layers, and we have services, topologies, performance, a lot of dimensions to the same data.
[13:36]You want, when you would like to ask questions of, um, what links, for example, are being used by, uh, by an OMS optical link, or, um, what, which LSPs I had in my network two days ago, which uses some specific optical devices?
[13:51]Many complex questions that you maybe would like to analyze, or even simple questions that you would like to consume by the different systems, you have, uh, you have a way now to, uh, uh, to use and to consume this information, because we have built an application layer on top of the data.
[14:07]And this application layer actually runs into the database, which is a, uh, relational database, but is able to answer questions of a multi-dimension questions, like the ones that I mentioned.
[14:21]And, let's take a look at this example. What if I want now to find if there are two LSPs, they share the same OMS, same optical link, and that maybe a problem, because I know that I've created those two LSPs for protection, but if this OMS link goes down, it means that I'll have no protection.
[14:48]How would I find that? Now, if you imagine running such a question on a database, that you have the information. But again, to take this information, it means that you have to run code by yourself.
[15:00]You have to take information on the LSPs, you have to get information on the OMSs, filter the information, and then get an answer to your question. We're able to get this question answered by simple question and simple query.
[15:20]Let's look at that. At the first, I get the link with a specific name, so I get now the first LSP. And then, I get the OMSs that these LSPs traversed through. Like, I do use the word downward, which is a keyword in SHQL.
[15:38]And it means that when I have the LSP now, and I use the word downward, it means that I get now down in the layers of the network, and I can specify the exact layer that I would like to get into.
[15:52]So, for example, now I use the word downward, then I get the link in the layers of OMSs. And I have now all the OMSs that are being used by this LSP. And I put it in an alias. Now I get the second one.
[16:06]And the way to get a second one is also by the name. And I get now also downward the OMSs and I get it in a different areas.
[16:16]Now I do a crossing between the two groups, and I find if there are OMSs that are in the first group and in the second group, in a simple query. And I get the answers right in, uh, in the, in the report table that I have here, which the same question can be answered not only in the application, but also in a REST API, that you can simply use and get, and get answer to that.
[16:42]Many use cases can be answered by the SHQL. Some can be, for example, layer relations. The one that I, that I just show, if I want to find all links that are being used by specific other link, fiber to layer three VPN, for example.
[16:57]Um, I can use time machine with SHQL, and I can simply put the day that I want to get back into. And all the information that I will get will be of a specific date.
[17:10]I can find shared risks, for example, if those two LSPs or other links in different layers, using same resources, then in a simple query, I can get answer to that.
[17:23]And I can also do root cause analysis by understanding, um, which services were impacted. And I can get reports, because one of the tools in, in SHQL is the ability not only to get specific information,
[17:39]but also to, uh, structure the information in a very specific way. So, for example, when you get the information, you get a default way of the report, like the columns with the information that is structured.
[17:49]And then you can decide in a simple query, which columns will appear, what will be the name of the column. So, in many cases, when you have a system, which was built to get some, um, files as inputs.
[18:00]So, you construct the files here. And those files can be inserted directly into tools, and you don't need to write again the code in order to adjust with the files that are exposed by SHQL.
[18:15]Um, there is a joint solution between NSO and HCO, which, um, comes back to the question of provisioning. Now, HCO is a tool for provisioning, but, um,
[18:31]when you look at the provisioning use cases, then there are use cases for provisioning of, um, optical devices, which is compared, I would say to layer three VPN, layer two VPN, this is, uh, fairly easy because they're all structured.
[18:47]Meaning, there are not many customizations between different customers when configuring optical devices. So, we have a solution in HCO to provision optical services across different domains, across different controllers.
[19:04]And, uh, that is using the same wizards in the UI and the same REST API. You don't need to, uh, adjust that to the specific vendor, but it works the same.
[19:15]But when it comes to IP services, and especially when it comes to, uh, uh, to run use case, for example, then we do use NSO for that.
[19:24]We have the NSO engine besides the HCO. And we understand that part of the complexity is to use all the programmability options of the NSO to provision the layer three VPN and the layer two VPN.
[19:37]And then we discover this information from the network and being able to visualize the services that were created by, uh, by NSO. And, uh, this is especially relevant in cases of a RON, for example, when we have the ZR pluggables in the routers.
[19:54]They're connecting to optical line systems by different vendors. And now we would like to provision a service, provision a link between the ZR pluggables over the optical line systems, being able to configure also the optical line systems, and then to visualize those links in the HCO.
[20:12]So for that purpose, we use NSO as an engine, and then we're able to do that in, in a click from the UI of the HCO.
[20:25]So, it is hidden from the users. They don't know that you use NSO for that. It's a simple UI in the HCO. You use the wizard, we configure that using NSO.
[20:34]Getting the information from the network and being able to, uh, to visualize the information. One of the questions that, uh, uh, we keep getting is about the adapters of the HCO.
[20:47]Meaning, how the HCO actually connect to the network. So, we do have adapters, which are equivalent to NSO nets, but they're different.
[20:57]They're getting a lot of information about the operational data, not only the configuration data. And those adapters are, um, being maintained and developed by the, by the BU, meaning by the engineering team of the HCO.
[21:11]And they are mainly purpose to work with controllers, like IP controllers, optical controllers, and, uh, and soon also microwave controllers. And in addition to that,
[21:24]Um, we are considering now extending this offering of the adapters, not only to controllers and NMS, by the way, but also to, uh, uh, to consider giving the option to create adapters, not only to the HCO team.
[21:40]This is in a very early stage, but we understand that, uh, um, there is a need actually to open the development of the adapters to other teams. So, every piece of information can be collected, no matter where information comes from.
[21:54]This is something that when we started actually the solution in Sedona time, we got this information also from different sources, like devices, using SNMP, CLI files, and, uh, and we would like to standardize the way of working with any source of information, and we expect that to, uh, um, to be developed in the coming, in coming year and being able to be available actually not only to the HCO team,
[22:21]but to other teams. Um, I would like to run a short demo and showing you the UI of the HCO, but not sure that it will work here.
[22:33]It seems not to, uh, to take the computer from the screen here. So I'm basically open for questions and, uh, this is the information about the HCO.
[22:52]Later today, there will be a session by, uh, Yona, my colleague, um, which is, uh, going to talk about the SHQL, and will elaborate about the SHQL options, showing you, uh, um, samples of the queries that can be running.
[23:23]Okay, then we will give up on that. And I'm here for the rest of the day, so any questions that, uh, would like to ask later, um, getting some more information about the use cases, about, um,
[23:36]SHQL as a language that can be used by developers. Um, myself and, uh, Yona is here and will be happy to answer questions. We have a question.
[24:06]So, you said before that the HCO has a PCE, and the PCE you can design a cross-domain, cross-layer service.
[24:16]Uh, how does it happen? Is, uh, does, uh, this PCE rely on lower level controller for for the PCE part or is it a full multi-layer, uh, PCE?
[24:28]So that really depends. It's a good question and it really depends if you would like to use the PCE offline, just to calculate a path and look at the path and take the information now to provisioning tools, tools or if you would like to actually provision a new service.
[24:43]Normally when we work in an environment of a multi-domain and you would like now to provision a new service, what we do is that we use the, the PCE of the controllers because they are the best in their domain and we're not aiming to replace the PCE in their domain.
[24:57]Um, yet we're able to, uh, use our PCE to find the stitching points between the domains and then actually, um, send each domain the request for the provisioning in the specific domain.
[25:10]Um, but in case of an offline operations, we're able to run our own PCE without using the PCE of the controllers. So, for example, if you want, if you want now to, uh, to plan a new service, and you would like to run, um, finding a path, getting all the constraints into the, uh, the pass preferences,
[25:25]you can use our tool to calculate a path. You don't need the PCE of the controllers. Okay, thank you. So, you support both black topology from PC perspective or white topology. So, you can rely on both.
[25:42]Okay. Thank you. Any other questions? We have a handy-dandy mic runner now, so, uh, so, I mean, can you talk a little bit about the some of the use cases that you, that you that we have customers deployed on this and the kinds of things they're doing?
[25:59]And when I talk to customers about this, I look at HCO, really creating intent, like the example you just used, and then NSO doing what NSO does, taking that intent and actually pushing down configs to you know, configuring end points, addressing the stitching points, those kinds of things.
[26:15]Um, so, you know, any other use cases you can share in terms of what practically customers are doing with the two tools? Yeah, well, um, it as I said, it's not only provisioning, because, uh, um, we, we take the provisioning to NSO, because NSO is best in provisioning.
[26:34]And we complement the provisioning with operational, operational data and operational use cases. Because for example, you would like to automate operation and now you would like to understand not only what the configuration status of the service is, but also what is the operational status of the service?
[26:52]What is the performance data of the information? Now if you want, if you work with a single domain and you don't need that information in HCO, that's fine. But if you work in, in an environment of multiple domains and you would like to get information on a service from separates, from multiple domains, then you need HCO to collect this information.
[27:12]And you would like to consume this information or the operational data of the service from a single point where you don't need now and, and it's crucially because we have the full path of the service end to end. We're able to tell you, for example, in the layer three VPN that is crossing different domains, what is the status of the service? What are all the end points of the service?
[27:28]Yeah. That's cool. And, and we can link to a demo when we post this, but, you know, one of the things I do find really cool is the ability to do the click down from, okay, a service died somewhere or something's wrong and be able to click all the way down from that all the way down to, you know, the lower level physical links and, and see what the issues are.
[27:46]It's, it's very, it's really powerful. We, we, I mean, we've talked a lot of the last couple of days about simplifying troubleshooting and and, uh, increasing visibility and, uh, analytics and HCO is actually a great example of, of how we're doing that. Yeah. Uh, last call.
[28:04]Moty is anxious to run around some more. Okay, then we will switch gears.
