Welcome everyone to another edition ofNetApp on NetApp for EMEA partners with us today we have Faisal Salam, Senior Storage Engineer and we will be discussing Amazon FSX for NetApp Ontap and its use cases at NetApp it. So I'll let you get started. Faisal. If anyone has any questions, please use the Q&A option or the chat if you have any more conversation. Thank you. Thanks, Ephra. Hello everyone. Welcome to this NetApp on NetApp session. My name is Faisal Salam. I'm a senior storage engineer with NetApp. It. My primary responsibilities are architecting storage solutions on premises and in the cloud. So as part of today's discussion, we will talk about our use cases of AWS FSx for ONTAP. Why we decided to use FSx. Compared with other cloud solutions, the key benefits cost analysis, observations and considerations. So at any point, feel free to interrupt me if you have any questions. Right. So I'd like to spend a few minutes to talk about our cloud journey. We started with NPS. That's NetApp Private Storage back in 2019. And this consisted of NetApp controllers deployed in a co-location. Um, this is also referred to as near cloud storage. So all of the compute in this design lived in AWS. In 2021 we then migrated many of our application workloads to Cloud Volumes Ontap when that became available both in AWS and Azure, and we continue to use CVO till date. Although the footprint has shrunk, we do have some presence in AWS as well as in Azure. Now, why did we make the switch from NPS to CVO? So NPS was on premises and we were responsible for all the storage and network deployments, the lifecycle management. ONTAP upgrades and all of that with CVO. With a few clicks, we were able to stand up instances very quickly and deploy virtual instances of ONTAP in the cloud because it's ONTAP. You know, we have all of the security and storage efficiency features thatcome with the operating system. And as this lives in AWS, we have the benefits of the cloud consumption model with CVO. However, it does require manual management, including licensing and software updates. And we still have to take care of all of those aspects. In 2022, AWS FSX for ONTAP was released and this was a game changer for us. This is a fully managed AWS service built on top of NetApp ONTAP being a AWS native service. FSx eliminates the administrative overhead of setting up and managing hardware, configuring, patching ONTAP software, orchestrating backups. All of that has been simplified and we will see that as part of today's talk. Now, FSx offers high availability and high durability, and there are multiple deployment options You could deploy it either in a single availability zone or in a multi-availability zone. When deployed in a single AZ, you get a file system that lives on a single node, and this is good only for production workloads. All of our FSX file systems are deployed using the Multi-az model, wherein we have two nodes, each deployed in a separate availability zone, and this is something we're using for all of our production workloads with high availability. Now, the way we have it set up, we have two SVMs per cluster, one for the lab and one for production. All of the EC2 instances, we attach a 100 gigabyte EBS as the root volume. Now application volumes are provisioned all on FSX andmounted on the servers. Uh, we have both Linux and Windows flavors. For Linux, we use both NFS and iSCSI provisioned Provision from the FSx file system for windows. iSCSI Luns are created and mounted onthe hosts. Now, to manage the AWS file system from on premises, you know that's made possible using the AWS Transit Gateway, right? So we're able to add it to our tools like DII and Blue XP. And we will look at how we are managing and monitoring our FSX file systems today. Now there's also data protection features that's available because this is ONTAP. You can leverage the snapshot policies. So we have the daily hourly and weekly snapshots configured. And all of our volumes have snapshot policies based on the retention requirements. We also use the AWS Backup service for disaster recovery In the event of a disaster in one region, we can easily recover the data in another region, mount the volumes on the servers andthereby restoring application functionality. How do we monitor FSx? We monitor volumes and aggregates using DII Data Infrastructure Insights, also known as Cloud Insights. And uh, we manage FSx using Blue XP. So all of our instances are added into Blue XP and DII forthe monitoring and alerting pieces. But that was a bit about our cloud journey. Were there any questions at this point? Any of the partners or customers share a similar journey from on premises? CVO FSx uh, are they all hybrid? So we are very much hybrid. And, uh, we think that's the way it's going to be for the foreseeable future. So we have presence on premises, uh, CVO uh, and FSx today. Nothing has been shared in the chat yet, but I'll let you know. Faisal. Thanks, Chris. Okay, so let's look at why FSx for ONTAP when there are many other solutions out there. Right? Uh, as the underlying technology is ONTAP, we can leverage all the superior features of this market leading operating system, and it also integrates seamlessly with all of our existing operational procedures and tools like ServiceNow. Um, so that's what we use for ticketing. And, uh, there's a web hook that's available within did that we leverage. And that's how we're sending us the alerts to ServiceNow. And the tickets are created forour operations team today. Now there's also the added benefit of integrating with existing AWS services like the Elastic Kubernetes Service and VMware cloud. Shouldthat be a requirement in your environments? Now one of the interesting features that I personally love is how easy it is to provision FSx. Uh, there are two options. There's the quick create option, uh, where it automatically uses most of the recommended best practices. Uh, and with very few inputs from the user, it goes out and creates the file system for you. If you want more configuration options, you could use the standard create where you get a lot more options that you could play with. Right now, programmatic access is available using AWS CLI. The ONTAP CLI, Ansible, Terraform to name a few tools. And what we've observed is that the overall cost reduces over a period of time when we compare with other solutions likeNPS, uh, AWS will chargeback for, uh, FSx and all of the other associated services thatwe consume in Amazon. Um. Now, like I mentioned, we monitor FSx using DII, but you could also use AWS CloudWatch. And we also have the options to send your system logs and audit logs, uh, to a collector like Splunk. Right. And from there you could spin up dashboards or send it to your ticketing tool and a lot of options that are available there once the data is ingested into Splunk, um, governance and compliance, we have NetApp classification. Uh, if you want to look at uh, PII data, this is available as a service in Blue XP. And for security you have the option to enable the AWS Key Management Service. And you also can integrate it with your directory services and for your database backups. We're leveraging SnapCenter. Um, talking about the migration options now, we migrated, uh, multiple applications from on premises to FSx, and we continue to work on that effort. And the tool of choice for us was SnapMirror because it is native to ONTAP, and it's very easy to set up with all the standard options that we have. You also could use NetApp Cloud Sync as one of the options to perform your migrations. Now I have many other solutions mentioned here like NetApp, Global File Cache, uh, AWS, Datasync, snowball, and all of these integrate well with FSx. Should that be a requirement forthe customer? Okay, before I move on, are there any questions? Chris. Yes, there actually is. Um, we have a question. AWS CloudWatch, CloudTrail and snowball do the NetApp service Cloud Insights Cloud Data Sense Cloud Sync work on AWS platform or are they just on the FSx platform? Uh, was that specific to Cloud Insights or are you asking about each of these services? Cloud Sync snowball data? Uh. I'm inferring from the question that asking about all of them, the CloudWatch, CloudTrail and snowball. Um. Okay. NetApp service cloud. Sorry. Go ahead. Right. Sowe're aware that these services are available for migratingdata from and into FSX. Uh, so I can gather more information about these specific services and get back to you. Uh, so, however, these are not the tools that we use for the migration. Like I mentioned, we use SnapMirror for migrations from on premises ONTAP into FSx. But these are options that you could leverage. And, uh, yeah, I can get back to you onthe specifics. I don't think we have any more questions. So I suppose you can continue on there. And if anyone has anything following up, please just put it in the Q&A or the chat. Yeah. So here we look at, uh, some of the key benefits from our experience with FSx. Uh, one of the key, uh, advantages is the ease with which you can deploy the file system, um, as well as monitor the storage related metrics directly from within the AWS FSx console. Now, of course, you could onboard the file systems into tools like DII, uh, to get, uh, more granular. But there are multiple options available, both on the, NetApp side aswell as on the AWS side. Um, now, because the performance tier uses SSD, uh, we can get really high throughput and low latency depending on what the requirements are. Um, and this is also extremely scalable. Now you can start with something like one terabyte. And if you want to increase the size of the file system, you could do that as like a percentage of, uh, your current size. Or you could put in an absolute value that you want your file system to be sized. And all of this can be done non-destructively. And how my team's doing this, uh, very frequently. And we've noticed, uh, no performance or functional issues whatsoever. So extremely scalable. Um, because this is ONTAP it comes with the benefits of thin provisioning, deduplication, and that enables us to optimize the resource allocations and avoid unnecessary expenses. Right. So there are workloads which we. Uh deduplicate. And within provisioning, uh, there's a lot of cost benefits that we get because, uh, this is FSx and, uh, because it runs data ONTAP. Um, another key benefit is multi-protocol access. So if you have a requirement wherein the same volume has to be accessed by multiple protocols like NFS and SMB, say home directories, uh, this is your place to go. Uh, FSx supports that. And you can also integrate FSX with Active Directory to manage user level access. Right. So you could, uh, simply do a CIFs, create and add that domain object into the add and set up domain authentication, the same exact same way as you do with your on premises disaster recovery. Uh, like I touched upon, we use the AWS Backup and Recovery Service, and that enables us to bring up the application services in the next available region in a very short span of time. Data protection? Uh SnapMirror. Uh, we have SnapCenter backing up thedatabases. We also have anti-ransomware turned on, uh, using tamper proof snapshots. Right. Uh, so all of our snapshots are locked with, uh, with a certain retention of, uh, you know, a certain number of days or hours, whatever the requirement may be. And because it's locked, uh, there's no way to delete those snapshots for the saturation. So, uh, there is anti-ransomware features that are available with FSx. Uh, SnapMirror. Uh, preserves the user and file level permissions because it is a block level replication. So in case of migrations, all you need to do is just break this SnapMirror mount the volumes on the server and you're pretty much good to go. And uh, another one of the key benefits is we don't manage the firmware and OS upgrades, right. AWS takes care of that for us. So those are some of the key benefits of from our experience withFSX. So are there any comments or questions aroundthe key benefits? Yeah. Faisal um, what advantages does FSX on NetApp ONTAP provide over our previous on premises solutions? Well, uh, one of the advantages that we've seen is,cost. Uh, that's one of the primary drivers. Mhm. Um, wherein we are able to have a cloud consumption model, wherein we can spin up provision and deprovision instances. Uh, it's not like on premises where you deploy a hardware and then probably you have drives, uh, sitting there spinning without having any workloads on them. Right. And,then, uh, it's very easy to scale, uh, you know, we,have options to go start from, uh, very low throughput or very low size. And, uh, we can scale up to a petabyte and several hundred, uh, Mbps, up to four GB of throughput. Right. So all of that scalability, uh, within a matter of minutes, uh, is one of another one of the key advantages that,we've noted when it comes to FSx. That's really interesting. Um, and I had a kind of follow up question, though you've already touched on it. Um, how does FSx impact, uh, scalability, which I know you already touched on, but encryption and upgrades. Correct. So like I mentioned, there's already native encryption thatcomes with AWS that's turned on bydefault. Uh, so that's, uh, taken care of. Uh, furthermore, what was the other question I missed that. It was just, um, how does it impact scalability, encryption and upgrades? Yeah. So upgrades, uh, areall taken care of by AWS. Now, they do give you the option to, uh, schedule those upgrades. So if you have any critical workloads running or there are certain times of day or week that you want to avoid, uh, you can let Amazon know, uh, and you could ship those upgrades that way, right. But for the most part, that's something that my teams are not involved in anymore because it's all it's a managed service. Uh, so it's all happening under the hood, and, uh, you know, it's all transparent to, uh, my team and to the application teams. Yeah, this is a good deal of flexibility there, which is very useful. Absolutely. Great. Thank you. Um, I believe we just got a question into the Q&A. When the file system size is increased simultaneously, do we need to increase the throughput capacity? Well, what we've been doing now, this very same scenario came up recently, right. So we had an aggregate that was almost full on the FSx. Uh, and so whatwe do is we look at the performance aspects of it too. So we did see performance, uh, pegged. Uh, so I would suggest looking at reviewing performance before increasing the throughput, although that's not a requirement. Uh, but definitely take a look at, uh, what your performance looks like on that file system, and then take a decision to increase the throughput if that's required. I think that's it for questions. I'm just typing in the answer, but I think you'll be able to continue there with your presentation. Thank you. Okay. Here's some of the cost analysis. Um, of course it's going to be based on, uh, the deployment type single AZ versus, uh, multiple availability zones. Um, with the performance or the SSD tier, we could start from one terabyte and scale up to one petabyte. I think that's what I last saw in the documentation. Um, the network throughput can scale from 128 MB to four GB. And, uh, you know, all of this is going to influence,the costs, All right. Um, now with performance, you have this concept of provisioned IOPs, which is three IOPs per gigabyte of SSD. That's the default. And this can be adjusted right. If you have higher performing workloads you can increase that based on your requirement. Um, now a really good technique, uh, to keep your costs low is to leverage the capacity pool storage or S3. Right. So data tiering for your cold data. Um, that's something we will take a look at. Um, you also have AWS backup service. So if depending on if you have that turned on, uh, that's,also going to add to your overall cost and to make, uh, kind of simplify things or to forecast your costs, you can use the FSx, uh, ONTAP pricing calculator. I've included a link over here And that that's something that, uh, we have leveraged. Um, now we do a comparison of, uh, the FSx file system with, uh, EBS and EFS. Those are the other services that's available from AWS. Now, in the case of FSx, uh, because of the native ONTAP features of thin provisioning, dedupe and compression, we have all of these space efficiency features that's available. Right. And these are all, uh, good for, uh, from your cost consideration perspective. Um, add to that, if you enable data tiering to S3, where your cold data is automatically moved, uh, that's also something that is going to help keep your costs low. Uh, the tiering you can control by setting in a default number of days. The data that's not accessed for a certain number of days will be considered cold. Then move to S3 and that's something that you could customize, um, with the FSx. Uh, SSD performance is shared by the entire file system and not specific volumes like EBS. And there's also the ability to create flex clones with FSX. Uh, so say you have some, uh, you know, development work going on wherein you need to quickly create a flex clone of last week's snapshot. Right? You could almost instantaneously spin up a flex clone in a few seconds, mount it on your compute, you know, take a look. Uh, and if you want to continue using that as a separate volume, you could very well split the clone. Or if you no longer require it, you could just. uh, deprovision that clone. Right. So that's also, uh, one of the features available, uh, because of data ONTAP. So I'm going to look at some of the observations from,NetApp IT. After over two plus years of using the product. Um, so we were able to start small as we started ingesting workloads in and uh, we've been scaling it, uh, as we move along, uh, we've enabled, uh, monitoring, um, and uh, we,do have ticketing that's happening for our volumes. Uh, and the aggregates, um, we're also using the SSD tier for improved performance. And like I mentioned, the ONTAP upgrades are taken care by AWS. That's a huge, uh, time, uh, saving measure formy for my team. Uh, Otherwise, you know, a lot of time would be taken up to, uh, plan for and execute the ONTAP upgrades. Like as we see with the On-premises or even CVO for that matter. Um, we're able to scale up file systems without any disruption whatsoever. And, um, we can also destroy these file systems if they're not in use. And, uh, there's also a high degree of automation, uh, using Ansible, AWS CloudFormation, Terraform. And, uh, migration from on premises to public cloud has been really easy because, uh, of SnapMirror and some of the other tools that we spoke about, um, I'll just, uh, show a snippet of how our footprint looks like. Uh, so we got a total of 14.5TB on the performance tier, which is on the SSD, of which only seven terabytes is used. Right. And you'd see that 11TB or cold data. And that's all been moved to our S3 or the capacity pool storage. So effectively you can see that we're only using like 50% of our high performing or our SSD. And that's primarily because of uh, one the,S3 tier, uh, the data tiering that we have in place and plus all of the storage efficiency and, uh, thin provisioning that's in place. All right. Uh, moving on. Some of the considerations, uh, the ONTAP release rollouts are controlled by AWS. And, uh, there are primarily two control planes, um, from AWS and from ONTAP. So AWS, uh, you could use the AWS FSX console. You also have the AWS CLI that's available. And on the ONTAP side you could connect it uh, like regular ONTAP. Um, you can create, uh, additional accounts. Uh, there's role based access that's available. Uh, you could create SVM level accounts, right? The same that we do with On-premises. So ONTAP CLI is one option and the other is. Uh Blue XP. Right. Uh, any questions so far? Yeah, just a quick question. Um, what accommodations were needed for the transition from CVO to FSx? Well.yeah. So, uh, some of the cost, uh, implications and some of the things like, uh, upgrades, like I mentioned. Right? Those are some things with CVO. I have to plan for the upgrades. Uh, do the change control. And we got to have dedicated resources that monitor us. And, you know, should we have issues, then we need to last with support and all of that. But with FSx, all of it is happening behind the scenes. Um, most of the times we don't even know that there's an upgrade that happened, uh, unless we talk to AWS. Uh, so, you know, the ease of management, um, and, uh, you know, mostly the,administrative, uh, ease with that comes with, uh, FSx. And cost is also another factor thatwe've noted, uh, when we compare it with, uh, with CVO. Okay. Um, and you already. Kind of thing. Yeah. No, you go ahead. Yeah, I was saying now, another consideration is there'sno FSx in Azure. So westill do CVO inAzure. Um, and uh, and yeah, that's,where all of our cloud workloads sit when it, when it comes to Azure instances. Okay. Um, and you already kind of touched on this, but just curious. Always a good question to ask. Um, were there any challenges or adjustments required during this migration? Um, well, really, um, because it was SnapMirror, you know, it was all, uh, really seamless. Now, of course, uh, there were some application workloads that had to be rearchitected, uh, right. Because there's a migration from on premises into the cloud. But other than that, for the most part, uh, because it'sall on tap. So on the storage layer, it was, uh, it was nice and easy for,the migrations and, uh, the post configurations, the monitoring, all of that was, uh, was easy or easy to do in our experience. Okay. Thank you. Um, some of the takeaways. Uh, FSx is, uh, fully managed service. You have, uh, multi-protocol support. Uh, like we touched upon. The same volume can be accessed over both, uh, NFS and SMB. Uh, should that be a requirement? Uh, you have high performance with, uh, SSD and, uh, the, uh, you have good options to scale the file system without, uh, any disruption. And, uh, because of the multi-availability zone deployment, you got high durability and, uh, and availability, um, data migrations aremade simple because of SnapMirror and some of the other tools that we talked about, uh, cost of optimization, uh, mainly because of the thin provisioning. Deduplication and compression. Uh, that'savailable out of the box with ONTAP. Uh, it integrates nicely with, uh, Active Directory, and it's got a pay as you go pricing. It's not an hourly pricing like we have for some of the other AWS services. Uh, so for example, if you're provisioning a file system of, uh, one terabyte, you're going to be charged for that one terabyte regardless of the time that you're going to be using it. Right. So that was a bit about the takeaways. Um, that'sall I really had. So I'll open it up for Q&A at this point. Um, so just giving people a minute to type in any questions they may have, I'll just continue to ask my own questions. Um, so going back to the FSx transition process, is this still going and are we continuing to move workloads to FSx? Right. So we did move, uh, multiple applications. Um, and uh, it is ongoing. We are going to have an upcoming, uh, round where we'll be moving further workloads into FSx. Soyes, it is, it is ongoing. But like I mentioned, it's, uh, we're going to be hybrid. We're going to have, uh, for the foreseeable future, wewill have workloads on premises, uh, in CVO and FSx, although that FSx footprint continues to grow. Interesting. And what do you feel are the key factors driving this transition? Uh, many of the things thatwe touched upon, Uh, likeI said, uh, data, the data tiering, uh, the, you know, the cost optimization that's available because of,S3, uh, the anti-ransomware features, uh, that are available, uh, and, uh, I think there are going to be more features that's still going to be available in,the upcoming releases. Um, and, uh, your ability to interface with other AWS services, many of whichwe looked at, right, xuh, a lot of that is made easier because, uh, now the storage lives in AWS, right? Um, so those are some of the, some of the things that,I can think about off the top of my head. But definitely there are a whole lot more advantages. Yeah, those new updates certainly sound exciting. Like they'll make everyone's lives a lot easier. Oh, yeah, it has for us.