Video: Virtual Workshop: | Duration: 2880s | Summary: Virtual Workshop: | Chapters: Welcome and Introduction (9.68s), Session Introduction (68.56s), Session Introduction (193.955s), Understanding Unstructured Data (302.6s), Legacy Backup Limitations (514.83s), Solution Architecture Overview (716.06s), Stateless Virtual Machine (829.845s), Cyber Resiliency Features (1091.93s), Backup Architecture Process (1266.985s), Data Recovery Process (1638.99s), Anomaly Detection (1901.475s), Guided Lab Walkthrough (2270.145s), Closing and Q&A (2677.1s)
Transcript for "Virtual Workshop:": Hello, everybody, and welcome. I'm gonna get started in just a couple minutes here, but really appreciate you taking the time on this lovely Wednesday. We were just talking about how busy it is. It's almost June. Time is flying. But, again, appreciate you, making the time today. If you're here to talk about unstructured data, you're in the right place, so stay on. And if you're not, stay anyway. You might learn something. You never know. Just bear with us as we as we wait a couple minutes past the hour. Awesome. Alright. We can go ahead and kick kick it off and knowing that people will filter in over the next three and four minutes. But, thankfully, they won't be missing too much. I'm just gonna go through, some of the housekeeping items. First off, my name is Hannah. I work on our programs team here, so I'll be helping with things on the back end when it comes to q and a or, yeah, any any action items from the actual experience today, but really excited to have you joining us. We have an awesome, content lined up for you today with Alpika. If you haven't heard her speak or lead a session, you are in for a treat. She's fantastic. She's one of our, you know, number one subject matter experts, with unstructured data. So really excited to get everyone's feedback. The big thing for today is going to be as much engagement and participation as possible. Now we're not crazy. We're not asking you to come off camera. We're not asking you to do anything like that. However, if you have questions, feedback, thoughts, you know, challenges even to the content, please share them in the chat or the q and a. Really, what good questions and thoughts are what make these sessions so worthwhile for everyone joining. So just keep that in mind. I understand not everyone can participate. Maybe you're just listening in, but if you can, it makes a big difference. So if you navigate to the right hand side of your screen, there's both the chat and then there's a docs and a q and a tab. Docs tab just has some resources for you. The chat is where you can ask questions or comments and as well as the q and a. We'll keep an eye on both. So, yeah, please do that if you can. I'm excited to see if we have, you know, a curious audience today, but at the same time, if you're just here to listen, that's great too. We have an hour, but we might end a little bit earlier than that. So we'll see how the day shakes out. But first, I'll kick it to Alpika to walk through some of the agenda for today, some of the presentation, and then don't worry, the guided lab experience will follow shortly after. So I'll kick it to you, Alpika. Thank you. Thank you so much, Hannah, for kicking kicking us off. And good morning, good afternoon, everyone, depending on where you're joining us from. Welcome to this rubric workshop session for your unstructured data. Like Hannah mentioned, if you're here for this session, amazing, you know, stay on. And if you're not, still stay on. You will learn a lot during this session as well. Now, my name is Alpika Singh, and I'm a Senior Technical Marketing Manager here at Rubrik. Been here almost four and a half years, and I, I look after our unstructured AI, SaaS, and database workloads. With me here is actually, Jeff. Jeff, he's our SE for unstructured data. He is helping me in the background to answer your questions. So, you know, we are prepared. While I'm talking, there are a lot of questions. Jeff is there, to help us out, answer all your questions. And he talks to our customers' prospects on a day to day basis. Right? So, don't be shy if you have very specific questions regarding your infrastructure. Don't be alarmed. We have the right set of people to answer your question. With that, we'll get into the agenda. I'll kick us off with some deep dive and some technical discussion, give you a bit of a preview as to what you will see in the guided lab Q and A. If there are a lot of questions, I can pause anytime, answer the question live, and then we'll leave you with the guided lab. Let's get started. First of all, to set the premise right, what is unstructured data? Unstructured data by definition is data that doesn't fit into your standard databases such as your SQL databases or Oracle databases. Best example, in my opinion, is your own desktop. It has files of different kinds, PDF, Word files. If you look at your workspace, Google Workspace or your M365, there are PowerPoint presentations of different sizes, Google Slides, Google Sheets. Pretty much any data that doesn't fit into a typical database is unstructured data. Every vertical has different type of unstructured data. If we look at student, like university or educational vertical, there are research papers, there are, IT papers, there are, your assignment papers. And from the teacher side, it could be your lessons and things like that. Everything, makes it unstructured data. Same thing with, patient portal. Right? Like your hospitals. They have, patient records, patient data, MRIs, scans, blood reports, so much so much data, and it's all unstructured in nature. With that being the case, you must be wondering why are we talking about it now. Right? So, data is just multiplying. It doesn't matter where, but in every industry, data is dramatically increasing. Also with AI, all the outputs that you get is all unstructured in nature. If you look at your holistic data, about 90% of the data is unstructured in nature. Now, when it comes to structured data, you always have a backup in place. For your SQL databases or for your Oracle databases, you have a backup recovery. The RPOs, everything in place, you know exactly what to do. But it's really important to have the same backup scenario even for your unstructured data because A, that this data is spread throughout. It could be On prem across object stores, it could be in different Clouds. You still need to secure this data because this data is fueling your own, organization. So, like, for example, like hospitals, they do need your patient records. Right? So you still need to safeguard it. And a lot of this data is sensitive in nature. About 25,000,000 sensitive files is what a typical organizations manage on a day to day basis. If you are larger organizations, there is a high probability that you have much more sensitive files. It's really important that you know where the sensitive data is and you protect this data the same way as you protect your structured data. What about backup options? It's not like there have not been any options so far because unstructured data is around for a very long time. Now, the legacy backups, the first option is essentially snap and replicate. Like the name says, you take a snapshot, from your primary storage, and then replicate to a secondary storage. Now, if you get hit by a ransomware, if you take the snapshot and you replicate to your secondary storage, your ransomware actually travels with it. Even if there is immutable data on your secondary side, you're still compromised because those credentials are same, both the systems are same. If the attacker has options to delete, it'll still be able to delete your secondary storage. It's not in a very logically air gapped. It's not a logically air gapped solution because they all live on the same network and infrastructure. The other option is NDMP. Now this was designed in the '90s where the dataset the largest dataset was five petabyte, and that was of the US Congress, library. Right? Now if you look at 100 or maybe tens to 100 terabyte, NDMP worked great. But now every organization or most of the organizations have terabyte scales of storage. Right? Now if that happens, NDMP becomes a very slow, solution with a lot of performance bottlenecks because there's a lot of data. Even if it's incremental data, there's a lot of data that needs to be backed up in a very short SLA window. So it it's not a viable option anymore. Of course, favorite option is archiving. Now, it's a complex management process. If you've had it, you know about it, you need to bring your tape online, figure out where your data is located, and it's not like you can just recover that one file. You have to probably recover that entire directory. That's a lot of work to be done if you just need to restore a single file. Net net, they all are very expensive and very inflexible options for any backup to have in this DNA each. Now this is where Rubrik NAS CloudDirect comes in play. What we do is that we provide you with data security and a very highly performant, solution for your unstructured data. It doesn't matter. You have hundreds of terabyte to tens of petabyte. We can scale from small to large. If if you have a growing dataset, essentially, you can grow with, Rubrik, Rubrik's solution. So what does it do? First off, it lets it scans, indexes, and move billions of files and petabytes of data from your source to your destination. It gives you 10x, the legacy speed, and then of course it provides you protection against ransomware as well. Let's dive in deeper in each one of them. Now the solution has three aspects to it. If you look at your left side, it actually lists a whole bunch of, source files, right, source, source systems. Now don't be alarmed if your, system is not listed here. This is not meant to be a comprehensive list. It's just for example purposes. Essentially, Rubrik has API integration with a lot of your NAS vendors out there. What benefits does API integration provides? First off, what we do is we take a snapshot, before we do any backup. So, and then we basically identify all the files that have changed using the snapshot and then move the data to the target and then do the deletion of the snapshot as well. If your system doesn't have API integration, then what happens? No problem. We can still read from the live file system. All good. We can also work with you to put steps in place to create the snapshot. We can then read from the snapshot and then the script can actually delete the snapshot also. We can automate that process as well, not to worry. The second component is the stateless virtual machine. Like the name states, it's stateless, so we don't store data. It only helps with the movement of data. Now, this stateless virtual machine lives in your datacenter. You essentially deploy it either in your Cloud or On prem environment. The Stateless Virtual Machine connects to the source and then scans, index, and move all of that data to the target. Now, from the target perspective, we do support your Cloud or your On prem NFS or object shares. Now, one thing to note is that Rubrik also has this offering of Rubrik Cloud Vault, which is our subscription based Cloud offering for storing your backups. So this provides you with that air gapped and immutable storage. That's also unsupported option, for storing your target, for storing your backups. Now, Rubrik Cloud Vault is also supported across AWS, Azure, and Google Cloud. If you have a favorite Cloud provider, you can choose that with RCV. Now, one thing to note is that as part of the scanning and the indexing process, we also calculate a lot of metadata. Now, this metadata is not your sensitive information or the content of the file, but when was the file created, who has access to it, what is the hash value for the file, and there's a reason why we use this. The indexing of the data helps us to search for files. If you just want to recover single file, you can use the index data to search and recover that file. There are other uses of metadata which I will get to in just a little bit. Yeah. We are able to scan a lot of data very quickly. How do we do that? We have customized NFS clients or object clients, essentially, that connects to your source. Depending on if you are using NFS, SMB, or Object Stores, we use the right client and the stateless virtual machine then spins up a lot of threads. What are these threads doing? They are essentially scanning, indexing, and moving the data. Each thread is independent of the other thread. If one thread is just scanning, it will only do the scanning. Once the scanning is complete, it'll pass it to an available thread that is going to index and that will be passed on to the copy portion of it. Now, we spin up a lot of threads to maximize the network output, network throughput. Now, we also are monitoring the latency on your primary storage. If the latency is not acceptable, it's greater than one hundred microseconds, I believe. If it's greater than that, we will stop creating more worker threads and we'll start, essentially, drilling or rather killing some of the worker threads down so that the latency is in the acceptable range. Each virtual machine is about five GBPS. If you have a lot of data, you can deploy multiple virtual machine in your data center, and, we can essentially parallelize, the backup process as well to meet, your performance requirements. Then the cyber resiliency aspect. As a platform, Rubrik is built on zero trust, so we never trust anything. You always verify. We we have built in MFA, RBAC encryption, for every data that is, in flight, the data is secure. Address encryption is your responsibility. We don't play in the address encryption portion of it, but any data that is in flight is always encrypted. Also the logically air gap protection. Now, like I mentioned, the Rubik Cloud Vault or RCV, if you are protecting your data with RCV, we guarantee the logically air gapped protection. If you are storing data in your datacenter or Cloud accounts, you are responsible for making sure that there are credentials that are isolated and it's not shared with multiple people. Next is a data threat analytics. Remember we talked about all the metadata and indexing. Rubrik also has a lot of inbuilt security applications to track your metadata and identify if there are any signs of encryption, ransomware, and also large changes happening in your infrastructure. It can help you identify the blast radius and identify a clean snapshot as well. We offer threat monitoring, which is proactive. After every backup, essentially, Rubrik will proactively scan that backup for any zero day or any indicators of compromise to make sure that your environment is safe. You can use these tools in conjunction with all the other security tools you have in place. The last part is the sensitive data monitoring. Now, what TrueBit does is that it scans your backup, and essentially look for different type of sensitive information. So if you have a credit card information, PII data, and if it has open access, Rubrik is able to identify exactly where your sensitive data lives, who has access to it, and if the access has changed over time. This way you can reduce your attack surface and meet your compliance requirements as well, especially if it comes to HIPAA or GDPR, making sure that the data is locked in. The validated sources. There are a lot of on prem Cloud and your object stores, which either we have API integrations with or we have validated them without API integration. Such as a whole bunch of stuff on prem, so QMLOS, VAST, GPFS, and even on Cloud side and then s three. We also have generic SMB, NFS, and s three S3 clients through which you can add your system. If your system is not listed here, don't be alarmed. You can just use the generic plugins for onboarding your sources. This keeps expanding, so you can talk to your sales rep or even look at Rubrik documentation to see most up to date validated sources. Now, looking at the high level architecture, how the backup process works. Now, there are three components. First is the customer datacenter or your Cloud tenant, which is where your source lives and the stateless virtual machine lives. The next one is the Rubrik Managed Control Plane, which is Rubrik Security Cloud and the NAS Cloud Direct Control Plane that has all your index databases and other things. The last aspect or the third bucket is your target. This can be the Rubrik Azure tenant, which is RCV, or your own Cloud account or your customer data center. Just depends on what you're using as your target. First step, the stateless virtual machine connects to your datacenter or your source, all the, essentially scans the file system and identifies what is the new data. Now, if it's your first time backup, all the data is identified as new and the backup happens. If it's not, then if it's your subsequent backup, say second or even two hundredth backup, the virtual machine will identify the data that has changed from the previous backup and just move that data. We give you this incremental forever backup, making sure that we only copy the new data and give you the efficient backup solution as well. Once the data is identified, it's moved to your Cloud account or your data center. Then all the metadata is sent to the control plane. Once that is completed, then the Virtual Machine commits the metadata and the job status to RSC or essentially Rubrik's Query Cloud. This is done on purpose. Unless and until the entire backup operation is completed, entire index data and all the metadata is committed in the database, we don't want to commit the job. To ensure that there is no data corruption and things like that, that's why we wait for everything to finish. Now, let's dive down a little bit deeper to understand what exactly, what are the different steps that happens during a backup process. I'm just going to break down the Stateless Virtual Machine on all the steps that actually happens. First is obviously a scan job. The Stateless Virtual Machine connects to the source NAS via API integration or just reads the live file system, then identifies and basically it calculates the metadata. Now, next step is to go talk to your NCD control plane, go talk to the index database, and then, essentially get the information for the previous back, backup. Right? And then Then it does the comparison. Previous snapshot versus your latest scan, and then it calculates all the diff. Once that diff is calculated, the stateless VM will then read the diff data, the data that actually needs to be backed up. Then will be the grouping or chunking. If the file size is less than 20 MB, they are grouped to form a single, file of 100 megabyte. And if they are larger than m 100 megabyte, they are, essentially chunked to be 100 megabyte, to be precise. Now once this grouping and chunking is completed, we go and then compress the data using l z four algorithm. And then we move the data to the destination, which can be your data center, Cloud account, or the Rubrik Azure tenant, or Google, or any tenant, or the RCV feature. Once that is completed, the metadata and metadata is committed to the index database and the process is complete. Recovery is literally just the opposite process. You as an admin go to Rubrik Security Cloud, identify what data needs to be recovered. It can be a single file or mass data, like it could be an entire share. It just depends on what you're doing. Once you select the snapshot from which you want to recover, everything gets in place. From the UI, a command is obviously sent to the stateless virtual machine to kick start the recovery process, and then the restore service gets into play. It then essentially talks to the index database and identifies all the list of files that needs to be updated, and also where this data is located in the backup storage. Then it creates a replan. What is where the data is, which data will be recovered first? It has literally a step by step plan for reading the data from your target. Once that is completed, the data is then restored in two phases. One of the phases is restoration of the small files that were grouped, and then, the other would be for the large files. If you're restoring to a source NAS that lives on prem, Essentially, first, the entire directory tree is restored. Next would be the large files, then would be the small files, and then the permissions are, restored on the recovered data. If it's a single file, then essentially you're just recovering that file and then applying the permissions that was there before. If it's an object store, then your large files or large objects are recovered with the metadata, and then the small objects are recovered with the metadata. Copy. Now this is a very cool feature that is offered by the solution. Essentially, Rubrik allows you to copy data from anywhere to anywhere, so your source NAS to your target NAS. It can be from Cloud to, on prem or on prem to Cloud or from one Cloud to the other. We support that movement of data. And, you know, if you think about it, it comes in really handy, if you want to train your AI. If your source or the most important critical data lives on your On prem and of course, all the training data actually needs to live in Cloud for training your AI agents. You can use Rubrik for moving your source data from On prem to any Cloud to make sure that your agents have the right set of data fairly quickly. The way it works is that the Stateless Virtual Machine connects to your source mass, same as the previous process. Backup process essentially scans the data, identifies the new data by comparing the previous snapshot or previous copy operation, calculates all the diff, and then reads the data from source and literally just moves it. Now, if this is a first time copy operation, of course, there won't be any diff. We will directly go to step four after step one. Essentially, you scan the data, there is no new data because it's the first copy. You move directly to reading that entire dataset and then moving it to wherever your target is. Then the job is committed, and then all the indexes are written to your index database. Now, we talked a lot about all the metadata calculation that happens during the backup process. What do we do with that data? What we do is anomaly detection. This is one of the application that is part of our Data Threat Analytics Suite. Then as the name suggests, we are looking for anomalies. What are anomalies? Things like massive deletion events, any encryption that happens. Now, if you're encrypting your data, we will obviously know. It is looking for encryption using, say, LogBit or any other malware that is prevalent out there. We look for those uncommon changes that happens in your environment. If you think about it, we are doing the backup, we have a snapshot chain for your backup. We are constantly comparing your current backup with your previous backup. We understand the usual rate of change, usual deletion. So we learn about your, environment over a period of time. Now if suddenly someone comes and just blows up your entire share, that is very unusual. Never has it has never happened. Or suddenly, if your data was not encrypted or if it was encrypted using certain key and, you know, everything changes, from the encryption perspective. We are looking for those changes that, usually ransomware attackers do to, you know, get your get to make sure that you're unable to recover, if if a cyber attack happens. So how does it work? Once the first step is obviously the entropy scan on your backup data. We scan the backup, look for any encryption, signs of encryption, and we send that information to Rubrik Security Cloud. The second phase is the anomaly detection part. So we compare your current backup's file metadata, so FMD, and your previous backup's file metadata, FMD again, and then we create a diff, file metadata. That diff then goes through our machine multistage machine learning model, and we are looking for anomalies. Right? So the machine learning model where you look for anomalies, identify if there are any anomalies. It will create the anomaly list, and then it is sent to anomaly detection UI which lives in Rubik's Security Cloud. Now, this anomaly list will identify the exact object or exact share where the anomalies have happened, when did it first start, what are the changes, if there are massive deletion events, updates, files being replaced, things like that is, these information is then sent to RSC and you can see that. The the entropy scan results, when it goes to Rubrik Security Cloud, there is a ransomware SSTable that lives on RSC. That table is updated and that undergoes again through a machine learning model to look for that encryption. If there are any encryptions found, we again send that information back to anomaly detection UI, and then you will see the collective information of all the anomalies as well as the encryption results. The last thing I just wanted to touch upon was data discovery and classification. So knowing where your sensitive data is is very, very, very important. As I mentioned earlier, it fuels your different workflows and it's important to know where it lives. Now for this feature currently, you do need to backup your data in RCV. And once the backup is completed, and, exocompute is launched, which scans your backup, and looks for any PII, PCI, HIPAA, or your custom policies. If you belong to a financial institution and you have different set of regulations, you can create custom policies that look for different type of sensitive information. The Exo compute will then run-in the background and identify where that data actually lives. And if there are any sensitive data violations found. Things like everyone has access to that sensitive data or the access has changed or everyone can go and modify or delete the data, things like that. If there are any such violations found, we do tell you if there are none, we will still tell you where your sensitive data lives. You can choose to move that data if it's not the intended location for that data. Once these results are displayed, after fourteen days, the data from the hot tier in our CV is moved to cheaper tier. Essentially, it could live in the hot tier or it can just go in the backup or archive tier. With that, what I wanted to do is show you exactly how the solution works, and what you will see as part of the guided lab. So, we have all sorts of topics. So the source onboarding, we have global SLA policies, how to create them, how to assign them, how do you do massive recovery operations, or how can you search for file and folder. We talk about anomaly detection, threat monitoring, threat hunting, and lastly, data discovery and classification as well. It's a very extensive, guided lab which you can take advantage of and learn from it. I'm just going to, go through the onboarding process just to show you what happens. You can actually use the right or left arrow key to move forward or back. So, you head to the right, inventory object and then simply go and click on add systems. Once you go there, you have option to select the site. Site is where your stateless virtual machine is deployed. You select that and then the system type. Now if you look here, we have so many systems listed. But what I wanted to show you was how we have the generic NFS, NFS v four, s three, SMB, all these app options available if your, system is not listed here. So you either use, if your system is listed here, you select that or use the generic option. Once you select that, right now I'm going to onboard an Isilon. I selected Isilon, put in your host name, credentials, and just click on Add. So, again, like, these admin credentials are not stored. We actually create, an admin account, with minimum sets of permissions that are required to complete the backup and recovery job. So once that admin account is created, the global admin account is deleted from memory or anywhere else. It's not stored anywhere with Rubik. So, once the system is added, Rubik automatically identifies all the namespaces available, all the shares, everything is available for you. Now, what is the next step? You obviously have to start protecting it. How do you protect? You need to identify what are the SLA requirements and then create SLA policy with that, requirement. So, just head to SLA domain, click on, SLA domain, and then, select NAS Cloud direct. Oh, yeah. We do offer this feature called retention lock. Essentially, if you turn this on, then no one can come and just simply shorten the duration of your backups. If if you genuinely want to reduce the backup window or change do anything change anything associated with this SLA domain, two preapproved admins have to call in Druid Security Cloud. They have their way of, identifying the users and making sure that they are the right people. And then Rubrik can make changes to this SLA domain. K? So this is a feature that is built in to make sure that attackers just don't come and, you know, expire your backups. K? So, you essentially define the backup frequency, how often you want to backup and how often for how long you want to retain the snapshot. Once you do that oh, yeah. You can actually have multiple backup frequency defined. So it could be every, x number of hours every day, every month, every week, depending on what you need for, any compliance reasons. You can define all of that backup frequency and the retention period in the same SLA. Then Rubrik will automatically take the backups on the day or every so often defined in SA domain. After that, you just select the NAS CloudDirect site and then give, and essentially select where you want to back up the, your source files. Right? Now for different backup frequency, like, if you want to store, for say for your daily backups, you can have a target location, which can be your Cloud or, for another frequency, say every month, the monthly backups can go in another backup or target location, or you can keep them the same. It just depends. But we give you the flexibility, to store different backups at different location as well. Once you do that, just hit create and that's it. You're good to go. And then you simply assign this SLA, to your shares. Now you can assign it to a single share or you can assign it to all your share, at once. That's an option too. But, you know, if you don't want to back up your entire, share, that's fine as well from us. So you just simply collect, select and then assign the SLA you just created, review other details, and you're good to go. So now Rubrik will automatically manage the entire data life cycle. The first backup to where it lives, when it should be deleted, the additional backups, everything will happen automatically. You don't have to do anything. If you want to recover, there are multiple recovery options. I won't go into that in the interest of time, but you can recover single file or entire share. It's a supported option from Rubrik. Okay. With that, I'm at the end of the, presentation. And then I see that all the questions I've been answering. Thank you so much, Jeff. Any last minute questions you all have, for me? Anything else you want me to cover? We have fifteen minutes. If not, I will give you some time back, and hope to see you in another session next time. Hannah, over to you. Hi. I'll be good. Do you wanna speak at all to the guided lab experience before I drop the link? Oh, I did cover that, so all good. Great. Alright. Yeah. You can drop the link. I went ahead and did that. I'm gonna go ahead and pin it. I'll unpin my other messages, which should be at the top of the box of the chat box. And, yeah, I guess, final thoughts, if there are any other questions from the group, now would be a great time to drop them in the q and a or the chat. I'll take a look there as well. Be on the lookout for a couple follow-up resources in the next twenty four hours, just to thank you for attending along with some of the documents that I said are in the tab. So don't feel like if you don't click them now, you never will. Also, in the right hand top right hand of your screen, you'll see something that says, don't miss forward. So, if you are familiar with Rubik, hopefully, you're familiar with forward at this point. We're making a big splash both with both in person and virtual experience. So just click on that link. Check it out if you haven't already. There is something for everyone with forward and certainly a further discussion on unstructured data along with much, much more. So please check that out if you're interested. I also just launched the survey. So if you have a couple minutes, we'd love to get your feedback on the session. We're always looking to improve. You know, we put these, workshops on, for you. Right? So if we're if there's areas that, I don't know, we can change up, if there's something that really spoke to you that went really well, please let us know. Yeah. That that would go a long way. So with that being said, maybe we'll hang around for a minute or two just to see if there's any final questions, anything happening with the guided lab. But if not, enjoy the labs. Again, would love the feedback and hope to see you soon. There's a lot more of these events, that we host pretty much on a weekly basis, in different subject matters. So, don't let it stop here. Hopefully, there's piqued your interest enough to maybe join another one, regarding one of our other products. So yeah. Okay. Awesome. Cool. Well, I think that we're probably good to drop, Alpika. Perfect. Yeah. Alright. Well, thank you everybody so much. Thank you, Alpika, for an awesome session. Until next time, Thank. you. See you.