FSx for ONTAP: Enterprise-grade unified storage for any application

All right, good afternoon, everyone. How's everyone doing? Good. Thank you for being here. Um my name is Punit Dean. I'm the Senior Director of Product Management at NetApp focused specifically on our storage services with um Amazon Web Services FSx for ONTAP. Uh I have a pleasure of having with me um Train from Gilead Sciences as well as Justin from Thomson Reuters today.

What we'll do is we'll dig into um FSx for ONTAP architecture. Uh we review some of the key use cases and workloads and then the most exciting part will have Train and Justin come on stage and share their experience um on how using FSx for ONTAP has been helping, you know, with their business objectives.

So, you know, it's fourth day at Re:Invent and you are still here um coming to the session. I'm pretty sure you are interested about storage, right? Cause storage forms the foundation of any workload that you're deploying on cloud designing, architecting, choosing picking the right storage is so fundamental to so many different kinds of workloads. Be it, you know, storage for enterprise applications like SQL Servers, Oracle, ERP CRM applications or storage just for general purpose files, shares. There's a lot of unstructured data out there. In fact, the majority of the data in the world is unstructured. Or if you're a developer, you're looking for storage uh for your Kubernetes platforms or you're designing SaaS applications, storage forms, the foundation choice of storage, the protocol that you use. How do you protect your data? How do you think about disaster recovery? All these are some of the fundamental decisions that you need to make to ensure that the application that you are designing, supporting, deploying, scaling has the right data infrastructure when customers move to cloud um especially from on prem or customers are scaling their applications in cloud. There are few fundamental reasons and objectives, you know, uh customers are thinking about the first one is really improving their reliability without really increasing the operational burden of managing infrastructure. That's a big reason we find our customers moving to cloud.

The other one of course is creating a security availability durability bar compared to what you can do that on prem um increasing the agility and businesses are moving fast today, right? And it and infrastructure cannot hold back the speed of the business. So our customers are really looking for that speed and agility of running, deploying scaling their applications very, very quickly. And that happens with automation. How can you make these infrastructures autonomous without having a lot of manual interference or manual management in place?

And lastly, as um Doctor Walker also mentioned CTO of AWS this morning cost is becoming a fundamental decision factor in infrastructure, right? It is um it is a nonfunctional yet a very critical requirement for most of our customers. So lowering the TCO is always on top of our mind for majority of our customers and that leads us to FSx for ONTAP.

What is it? Right. It in a nutshell, FSx for ONTAP is, is NetApp’s ONTAP software now delivered as a fully managed highly scalable efficient service from AWS. It's a native AWS service. AWS sells, it supports it, deploys it, you have it globally available in all the regions. And the question, this question is what is ONTAP, right? Um I'm sure folks who have uh management been managing infrastructure on prem know a lot about ONTAP, but this is one of the world's leading data management platform, the most proven data management platform on prem. And for the last two years, it's been available as an Amazon AWS service.

It's a fully managed petabyte scale storage service which provides multi protocol storage. In fact, FSx for ONTAP is the only unified storage service available in AWS portfolio. There are multiple ways to manage it's fully integrated into AWS console, AWS APIs. Of course, all the ONTAP APIs which our customers have loved for decades are also available to take advantage of the advanced functionality. Uh it gives you the flexibility of deploying storage depending on your applications, availability and durability requirements from a single AZ to a multi AZ configuration without having to do the work of managing that availability across um across AZs.

I talked about the range of applications that you have to support right from, let's say, low performing file shares to very high performance software build environments to extremely high performance HPC or, or AI/ML applications. You have the flexibility to adapt performance based on the requirements of your applications.

Data protection is is on top of our minds, especially with, with what's happening with ransomware and all the attacks that we are seeing, protecting data uh is fundamental um to any application and ONTAP provides a lot of robust features to ensure that your data is protected within an AZ across AZs across regions between on prem and cloud pretty much everywhere. And lastly, as I said, we bring a ton of new, a ton of features to optimize the total cost of ownership and reduce the data footprint so that you are storing your data with the maximum possible efficiency.

Let's dig in a little bit into how FSx for ONTAP works at a very, very high level. Um we have two storage servers uh running the ONTAP data platform software uh providing access a secure multi protocol access through SMB NFS and pNFS for the unstructured data as well as through iSCSI um using um uh for block storage. In fact, the unstructured data besides NFS SMB can also now be accessed through S3 APIs for the applications that are coded to take advantage of an object storage platform. The same files can now be exposed as objects you could deploy um uh you could deploy these two storage servers either within a single AZ. So you still have high availability within an AZ or these two storage servers can be deployed across AZs so that if you can tolerate a full AZ failure without losing access to your data and your services. And by the way, all this is done through the automation of AWS, you don't have to deal with architecting a multi AZ storage. It's already taken care of when it comes to storing the data and giving the right performance levels.

We take advantage of multiple intelligent features like memory caching. So we take advantage of the memory in the instance as well as NVMe storage attached to that instance. We call it instant storage to accelerate application performance for reads and writes by caching that data as the data lands um on storage. It, it's the active data sits on an SSD high performance SSD tier. But we also all know that majority of the data um that we have in our environments, it's not actively used. So there is no point keeping all that data on an expensive SSD tier and what FSx ONTAP does is it has a very transparent tier technology where the inactive blocks again, based on the policies that you pick based on the workload requirements the inactive blocks of the data are tiered to a lower cost capacity pool. And not only that as the data is coming in or as data is at rest, we ensure to compress dedup use compaction to really reduce the footprint of the data so that you are maximizing the return of every penny that you're paying to, to store that data.

Moving on. I was talking earlier about how critical is um data management, data protection. And we bring a lot of new advanced features to FSx ONTAP. Uh that help you protect your data uh very rapidly. Um point in time um protection is becoming more and more critical for our customers where if there is a data corruption or there's a ransomware attack, you wanna go back to how your data looked 10 minutes ago, 10 hours ago, 10 days ago, right? And you can do that instantaneously without any impact to application performance as well as without any additional cost because our snapshots are thin snapshots and instantaneous snapshots.

Similarly, if you need to make copies of your data, we see that in software build environments quite often where you have multiple environments but needing access to the same data, but they want to have their own copy. So rather than making, let's say a copy of a 10 terabyte volume 10 times paying for 100 terabytes of data. We make thin clones just like that without paying a penny that's pretty powerful.

We also introduced um earlier this year. Um our compliance technology called SnapLock, uh Snap um SnapLock where you could write the data for meeting regulatory compliance needs so that once it's written, it cannot be touched. We have the most robust compliance technology um that is now available with FSx ONTAP to ensure that any of your critical data um can be met with the compliance and regulatory requirements.

When we talk about cross-region replication, cross uh uh replication between um an on prem to cloud or cloud to um on prem disaster recovery capability is really enhanced by our built in SnapMirror capability. So the whole, the whole data protection suite really enriches our the experience that our customers are getting without having to architect uh for those capabilities. Um on top of what is available natively in some of the other storage services.

And you know, the the best way to take advantage of storage is to ensure that it's available accessible integrated with a wide range of AWS services. So um we have the, the FSx for ONTAP is fully integrated with monitoring services like CloudWatch CloudTrail. It's integrated with AWS Backup. Uh for you to automatically takes uh the backups and schedule the backups. Um it has full integration with EKS Palo cluster kubernetes service and it's also accessible from wide range of compute resources like ec2, of course, um EKS for Kubernetes platforms um RedHat OpenShift platform um as well as um other AWS services like WorkSpaces. You know, you need file shares for the VDI um users, right? It's it's there for you to have a central NFS SMB file shares accessible for Windows clients, Mac clients, Linux clients consolidate all that storage into one place. Uh you could also utilize um NFS shares for example, with SageMaker notebooks um to, to, to provide shared storage for your data scientists.

And lastly, uh from a management perspective, uh there are multiple range of tools APIs available both within the AWS ecosystem, whether it's AWS CLI API, as well as the ONTAP API, ONTAP CLI so that you have the breadth and the depth needed to take advantage of all the automation capabilities that you can build uh to get that agility and speed that businesses need.

Let's dig into performance a little bit now so far. Um until about um last week, um our standard or scale up architecture where i just explained, you have two, you know, a single pair of HA servers. Um we were able to drive about four gigabytes per second throughput, uh about 160K IOPS at sub millisecond latency and about 192 terabytes in that SST pool and our customers have been utilizing this very happily, very satisfied across a range of applications.

However, some of our larger customers have also been um you know, running applications that require much higher scalability um to process large amounts of data with a large amount of compute capacity. And this week, uh we're very happy to announce our scale out architecture that really takes the performance to the next level. So from about four gigabytes per second throughput that we were able to achieve now, we can get to 36 gigabytes per second, nine times increase in performance capability that we have made available this week with AWS.

On the right side, you can get from about one gigabyte now, more than six gigabytes per second, right throughput, more than a million IOPS and more than a terabyte of SSD storage capability all within a single cluster. And how we do it is we utilize our scale out architecture that has been tested and proven in an on prem large on prem install base for years. And we have brought it now to FSx ONTAP as a service up to six pairs of highly available storage servers work together in tandem as a single storage cluster.

We utilize a technology called FlexGroup, which if you've been using uh FSx ONTAP, uh you already had a FlexGroup within a single node pair and now the same FlexGroup can be extended across six node pairs to give you the aggregate capacity, the aggregate performance across all those six pairs working together. This architecture um is it is gonna be extremely helpful for a wide number of use cases

Whether our customers are using high performance computing, electronic design automation, um media entertainment for vfx post production, we can bring a lot more artists now in a single storage solution and get the bandwidth and the throughput that they need um seismic analysis in oil and gas as well as uh for machine learning training where you need to read the data sets as quickly as possible. So that your compute, the GPU capacity is not waiting on the data from the storage.

We think these capabilities are gonna extend powerful value proposition for our customers who are trying to make their decisions much faster without really waiting on the technology.

We spoke about um use case like EDA right. Some of our largest customers are from the semiconductor industry and they've been pushing us to extend the value of FSx. Um so that the designs, you know, all the chips that we are using in our phones or computers, you know are in the cloud, they can be designed much faster at a lower cost. And this is an example of how our customers in the EDA space um are utilizing FSx for ONTAP um to, to um to, to build uh to uh design their build environments for um for the EDA applications.

A lot of our customers are hybrid customers. Um they want to ensure that they have the data uh not only in cloud but also on prem and they also want to ensure that there is a highly efficient data mobility solution available for them so that they can get the data where they need to be.

We all live in distributed teams spread across the globe today, right? And we need our developers, our designers to have access to the data very, very quickly, very efficiently. And a lot of our customers are utilizing a technology called FlexCache which provides a read write, cached copy of the data. Your source data could be on prem on NetApp, it could be in AWS cloud, it could be in one region and you can provide a very highly efficient cache of that data, you know, within minutes to your developers across the globe.

Our customers find that not only helping you to accelerate the development pipelines but also save on the cost because of the efficiency on how quickly and how efficiently the data is shared. And now we scale out these customers are able to fire up their jobs on multiple, you know, multiple compute instances and ensure that they will have the scalability of the storage at the data level to match the the throughput and the IOPS and the performance needs that the compute clusters would need.

Customers like Arm, we've been working for a long time. They're really excited to put their hands on scale out. And more than them, we are excited to make sure that we have been able to bring this capability into their hands so that they can design the chips much faster and you know, we can run our iPhones more faster too.

Let's switch the gear a little bit more um towards block storage. I said earlier FSx for ONTAP is the only unified storage service available in AWS uh portfolio and it provides block storage using iSCSI protocol for the applications that do need the high performance low latency, um small block IO while needing the high availability, the high durability um for those business and mission critical applications, very common pattern where customers want shared storage and not just shared storage but storage that is also available across AZs.

See if one AZ fails another AZ can take over without actually doing the operational um investment to make that happen performance and scalability are of course a key concern for these larger database environments. For example, our customers wanna create multiple copies of of the database environments. The customers also want cross-region replication again without doing the administrative work. And lastly, the block storage needs to be cost efficient. This has been tough to do in some of the native cloud storage block storage environments. But with FSx, you know, with a single storage solution. Our customers have been able to take advantage of all these capabilities for the applications that need block storage.

And here's one example for the database environments, a very common pattern we see for making sure the database is highly available across two AZs. And a lot of our customers have been using an application level replication or a third party replication tool which not only adds cost, it also adds complexity and issues with scalability. And now with FSx ONTAP using a multi AZ file system, you could have a block storage which can be shared within a cluster environment and it's available across AZs, for example, in a SQL Server environment.

Now, you don't need to use an enterprise licensing. You could deploy a SQL Server with standard licensing, reduce the cost, lower the number of cores that you need in the R2 instance and still able to get that cross um AZ availability to ensure that the storage is there for the most demanding applications that need the most uptime.

Here's one example, S&P Global, a big customer of ours utilizing FSx for ONTAP for their SQL Server environments. Hundreds of databases um hosted in a fail over cluster environment um with very aggressive RPO and RTA goals being met by moving from another block storage solution to FSx ONTAP.

The similar benefits extend to some of the other critical applications as well. Think SAP HANA one of our large um customers, they had very aggressive RPO RTO goals, a very large HANA environment that they needed to bring to AWS and, and, and FSx for ONTAP was the only solution that could actually help them meet very aggressive, uh aggressive goals.

They were able to, you know, bring the whole environment together, architect, the whole environment and bring up the whole environment within 25 minutes. It's a nine terabyte HANA database. That's that's the kind of um you know, speed that these customers are looking for and FSx for ONTAP helps them achieve these highly aggressive SLAs.

They were also able to take advantage of a single storage solution um providing flexibility of single AZ and multi AZ um for different HANA data sets. So for example, you they put um the HANA databases on a single AZ file system while the shared files live on a multi AZ file system. Again taking care of all the availability that is um that is needed yet, not, you know, putting that burden on the customer's hand to architect for that and still being able to take advantages of all the data protection features, the data efficiency features to bring down their cost. Why help them meet their very high SLA requirements.

VMware is another popular use case. You know, a lot of our customers utilize VMware vSphere for their on prem applications. And as these customers are moving to AWS. They are really looking for that simplicity without having to learn adapt to a new technology. They wanna extend that skill set, extend that simplicity of architecture to AWS and VMware Cloud on AWS provides that.

And not only that when you combine that with an enterprise storage, which customers have been using, you know, for decades now on prem and now that same storage FSx or ONTAP is available as an external data store with VMC. That's a very powerful combination to help our customers not only simplify that migration of VMA workload from on prem to to um to AWS at the same time, really helped them with lowering the total cost of ownership because now you can scale your compute and storage independently.

This has not been possible in the past where VMC architecture was just based on internal storage using VVOLs. And if you do not have um if you have large storage requirements but do not have large computer memory requirements, you needed to add a lot more host that added the cost. But now you could, you know, utilize a disaggregated compute and storage environment, especially now with the M7id.isc class nodes that uh we and we just announced this week at re:Invent.

It's a powerful combination to, to have uh a highly available architecture on a single or a multi AZ um as well as utilize all the data protection capabilities. um that we extend to a VMC environment. Just this week, we announced a preview of how you can take advantage of our application consistent snapshots for SQL and Oracle running on VMC infrastructure.

We're also running a preview of NetApp disaster recovery as a service for customers to easily run ADR environment um as a managed service from on prem to cloud or on prem to on prem and all that really, as I said earlier results into uh a highly uh compelling case for lowering the cost as well.

Here's an example, a real world example of a large manufacturing organization looking to bring about 2000 VMs to AWS uh with about uh you know, a petabyte of storage. And rather than going on a purely hyper converged vVols architecture, this customer was able to split their virtual machine um footprint uh and utilize an external storage combination with FSx for ONTAP and lower their overall cost.

I mean the savings between more than $2 million savings across three years, 2220 to 50% lower TCO that's net savings. These customers can now put to critical strategic areas. I know of their businesses that require that kind of investment, big, big value for our customers.

When we talk about developers. Um a lot of our developer community, the builder community is using EKS and the similar benefits now extend towards um Kubernetes environments too. A very common pattern. We see customers wanting to use multi AZ file systems um that have the high performance um that have a rich data protection features. They're simple to use in their builder environment because these developers do not have a whole lot of storage background, they just want to consume storage and with a Trident CSI driver um with our data protection features, we bring that simplicity and data protection capabilities to EKS environments.

Um one of our customers in Fortinet are now using E uh EKS plus FSx combination to process over 40 million messages a day at a very high speed, lower cost that has not been possible with some other storage solutions.

And how can I leave AI out? Have you not heard enough about AI this week? But you know, it it has been um a transformational and I think one of the most important innovations of our lifetime, but to unlock the true potential of Gen AI, you can only do that once you connect through your enterprise data in a secure fashion, you know, um in an easy fashion and that's what we are trying to do.

Um a lot of our customers are also struggling with the data mobility problem. Should I bring the model to the data or should I bring the data to the model and FSx for ONTAP along with our FlexCache technology helps you in both, right? uh really simplifying connecting the two together we're showcasing uh been showcasing a demo uh in our booth on how to connect Bedrock with FSx ONTAP to uh to deploy um um uh an example application like uh augmented chatbot utilizing NLP techniques so that you could ask questions. You can summarize your documents based on the data that you already have.

The multi protocol nature of the product really helps you to deploy multiple vector databases. For example, whether those vector databases require NFS storage object storage block storage, not only you can consolidate your data, but now you could run your vector databases on the same storage. You could protect the whole thing together. You could go back using snapshots, you know to you, you can go back point in time, you can develop a consistent disaster recovery strategy for your Gen AI application infrastructure.

Not only that you can consolidate different pipelines, different different pipelines, whether it's preprocessing, whether it's training, whether it's modern deployment all on a single storage solution and that is very powerful our customers, right? Not only helps with the simplest city of storage, choice storage management, but also extends the data protection and TCO advantages.

You know, across all different kind of applications to summarize. I would say regardless of whether you are, you know, looking at storage for general purpose files, shares storage for your databases, ERP CRM applications, software built environments or utilize or you're thinking of storage for Kubernetes environments, FSx for ONTAP provides you a very versatile, the only unified storage service available in AWS.

If you still have the energy uh by tomorrow, uh I would highly recommend to go um visit um um Venetian tomorrow morning at nine o'clock a w AWS is running a hands on lab uh where you can firsthand experience the scale out capabilities that we made available this week. So please go ahead and get your hands dirty and see how fast you can run your storage with the new architectural benefits we bring to the table and with that, let's shift the focus to the most exciting part of, of this session today.

Please welcome Trin. Welcome Justin on stage. Let's give them a big hand. You all right. Thank you for being here. Um uh Justin Tr uh why don't we get started with maybe introducing yourself what you do with, you know, introduce your company. Um and then also touch base a little bit on what were some of the problems that, you know, that you were trying to solve? And you know, what was, what were the things that you were struggling with uh that led you to look at AWS as, as your destination and especially FSx as, as a destination for storage. So let's, let's start with Justin.

You sure. Um hi, everyone, Justin Wright. Uh I head up a platform engineering for Thomson Reuters. I've been there for a little over 18 years now and Thompson Reuters, if you're not familiar with is a, a content driven technology company. So we sell software and services to professionals. Um and we also have a media news arm that um you may be familiar with.

Um so we, we've been on a cloud journey for a number of years. Uh we're about 85% of our workloads are in public cloud today. Um and we, we've really leveraged NetApp and FSx specifically is in some of our data center exit and workload migrations that we've been doing over the last couple of years where we've been using NetApp on our in our own data centers on premise and we've really become accustomed to the performance and throughput and speed and enterprise view that FSx or that NetApp has delivered to us now available in FSx.

So thank you Justin. So thank you for having me here today. It's an honor to be part of this conversation. My name is Trent and I manage Torch infrastructure at Gil Science. Science is a research based pharmaceutical company that has been pursued and achieved many breakthroughs in medicines for over 35 years. Our primary goal is to create a healthy world for all people.

Gilly is committed to advancing innovative medicines to help prevent and treat life threatening disease such as HIV, hepatitis, COVID-19 and cancer. Through our pioneering access and health factory programs, we were able to make our medicines available to millions of less fortunate community across the world and, and, and you guys have done amazing work, especially in the last pandemic we have seen. So, thank you.

Um and of course, I, I remember you guys when we have to file our taxes. So, but just to talk a little bit about some of the core challenges, why were you looking to exit the data centers? Uh what were the challenges or objectives that you know, led you to look at uh at, at cloud adoption.

Um so we started our enterprise cloud journey uh in uh mid to late 2015. Um we actually started with a cloud first kind of uh uh build new uh approach for a few years. And we became very um very much aware and a fan of like cloud native design, cloud native architectures, et cetera.

Um and then in 2018, we divested half of our company um uh our finance and risk business, which is now known as Refinitiv and l a company. And as part of that dives, we divested nine data centers and we decided to take that as an opportunity to really, you know, accelerate um the movement of workloads into public cloud.

Um and start with our first kind of migration or lift and shift if you want to use that term uh approach for, for a number of workloads, i think it was about 10,000 assets at the time. Um and uh uh we, we've finished that up. Um in 2021. And at that point, we were about 60% in the public cloud. And we thought, well, what are we gonna do with the rest of it?

Um and so we started uh exiting our last company owned data centers um in Egan, Minnesota uh that we're going to be fully out of by the end of next year.

Got it. And Trin, what were some of your uh business challenges or objectives that led you Gilet to think about cloud.

So if you manage your own data center, you can understand that infrastructure is a limiting factor. Um you only provision your environment just enough to meet your business needs plus additional buffer space to support their organic growth. You want to, you don't want to buy too much hardware because then it creates the unnecessary waste.

Um you want to get the big bang for your investment, but the problem with provisioning infrastructure just to meet your business requirement just for right now is that you don't have that flexibility to be agile. So the cloud kind of provide a path forward to that infrastructure limitations. It helps us remove unnecessary steps and it helps to streamline our process to increase productivity.

Um also one beautiful things about the cloud is the amount of resources that you can easily tap into. A TOS today has over 200 services that in a wide range of technology that is a lot of lego building blocks that you can add in your toolbox. So the more tools means the more capabilities, productivity and the ability to enable and support your business.

At the same time, it allowed us to be more agile and responsive to our business needs.

Totally. Totally. Let's go back to the on prem. Could you describe a little bit about what environment you said existing NetApp customer? Could you provide a little bit more detail on, you know, what kind of infrastructure you have there? And what kind of applications are you running on? Prem and, and then talk about your journey to move those to to AWS.

Yeah, so we have um we have hundreds of different products and applications. I mean, our, our on premise estate was pretty heavy, you know, oracle, pretty heavy, linux pretty heavy, you know, i guess you would say so a kind of architecture, design patterns and, and such um cloud as Trin mentioned, i mean, with the, you know, elasticity and the ability to get infrastructure on demand and everything has been hugely valuable for us.

Um we do sell tax products, so tax peak season um you know, other different peak seasons where we need capacity very quickly. Um and so as part of and, and obviously, we were a big NetApp customer and still are.

Um but as part of our migrations, what's been really helpful for us is to actually use um you know, the on tap software solution to help move data uh and migrate data. Um you know, a ahead of kind of that go live and that cut that uh rollover cut over if you will.

Um that's been extremely helpful. And how was that experience in utilizing a zume snap mirror to, to migrate that data?

Uh it's worked great. I mean, um not only are we familiar and very comfortable with snap mirror, but um it really helped accelerate some of the the migrations of systems where we were trying to do, you know, an incremental or basically an initial copy and then an incremental backup.

Um and, and a copy of, you know, right before we actually cut customers over. Um and uh yeah, it's worked, it's worked really well.

That's good to know Trent. How about your side.

So, one of the challenge for us was we going through this cloud migrations journey is that when we deploy new applications, you have that flexibility because we're highly regulated. It's very hard. It gets a little bit more tricky to move legacy applications from on premises to AWS refactoring and repla consumes a lot of time.

At the same time, it introduces risks and disruptions to a business. Therefore, with snap mirror and we've been NetApp customers for many years. So snap mirror has kind of helped us doing our migrations because one of the problems we have is that when your data is being shared between users and various applications to multiple protocol there wasn't a solutions on AWS that will help us allow that move that data because there wasn't a service on AWS that support multi protocol a year or two ago.

And so that becomes a challenge. Yes, you can lift and shift your compute to the cloud and leave your storage on prem or you can be a little bit creative and move your data to S3 and play it place the file gateway in the front as an example in some use cases that may work fine and other use cases, it may not work so fine because at the end of the day, file gateway is a virtual appliances with limited resources.

Will it meet your performance requirement? Um will it meet your high availability requirements? And does the cost architecture of that design make sense for you?

Right. So at the same time, if you have PCs and Mac user, as you're in computing, your Mac user will have will experience issues performance latencies and other problems because file gateways does not support SMB on Mac.

So for us, it moving using on to a since we are so experienced at having it, we have experience using it for so many years. Migrating to FSXn was easy is st and it has very little to no impact to our business. And like what Justin says, we don't have to learn a new technology and we are also able to take advantage of all the features and functionalities that we love would on tap on FSXn.

So the choice of storage that was very important, you know, even before using some storage, doing some homework on what storage is the right fit for the applications. I'm guessing that's, that's your message too, right. That's correct.

Do the homework in terms of the homework and when you do the, you will probably find out that depends on your use cases. FSX is probably more cost-effective compared to other storage services on AWS today, like EBS EFS or FSX for Windows, for example. And have you been able to utilize the compression and the tier benefits? That's correct.

And like everything you said, um we use a BS as extensions to our on prem data center as part of off site data protections and disaster recovery, Justin on your side. Were there other storage alternatives that you were looking for or was it a slam dunk decision that given you were a NetApp customer on prem? And, and FSX was a natural choice.

Uh so it, I mean, when we started out kind of 2015, 16, um it wasn't really an option yet. I mean, so we do have other workloads that are using more traditional AWS storage of offerings. Um but when NetApp became available via the AWS control plane via FS x, um it was very intriguing for us and, and we started looking at it.

Um you know, because we can use cloud formation. We can use infrastructures code all the best practices for, you know, cloud native design are there.

And the workloads that we were looking to, you know, at that point in time migrate required the level of throughput performance, low latency that NetApp can deliver and has done a good job for on prem. And we decided to give it a shot and we were getting the same, you know, if not better performance and throughput with FSx in AWS. And it's been extremely helpful for a lot of workloads that we've migrated the last couple of years.

When we were talking yesterday, Justin, you were also talking about the fact that, you know, the advantages of a fully managed solution versus having to manage the storage yourself. Could you, could you talk a bit a little bit on how it does help you?

Yeah. And so, you know, I'm sure a lot of folks here in the room, um you know, appreciate the managed service offerings that cloud provides. And, you know, we were looking at, you know, for our initial cloud migrations, like how can we be really smart and techno uh talented people and employees at, at Thompson Reuters. And they're working on a lot of things that maybe aren't quite differentiating to value for our customers. So we really wanted to take advantage of the managed service offering and um you know, refocus some of our, you know, talents to capabilities that are more visible in our products and our application. Um and so, it simplifies our support, simplifies kind of the, the running and the overhead of managing those things. Um so we get the value without having to, you know, install and run and support it ourselves.

Totally trent how far gilliard is that, has that been um a value driver for not having to manage storage?

Well, absolutely. I mean, we even some of the automations that we put in, you know, with NetApp on, you know, on prem, we used to have our operations team alerts monitors in place to manually goes in and grow NetApp volumes. When we reach a certain percentage. For example, we're able to move away from that with using lambdas and you know, watching the CloudWatch alerts trigger lambda functions to automatically grow whether it is the SSDy performance pool. Um that way we kind of remove and remove that operations, remove those unnecessary steps that help us be more productive.

So you are monitoring the storage utilization and at a certain point, you're firing a lambda function to automatically throw the storage, correct. That's pretty powerful, you know, removes the operations team, you know, from this whole process and makes the infrastructure a lot more autonomous to adapt to your changing requirements.

Talk a little bit Justin about the the agility, you know, you spoke about automation, how are you using automation and what kind of benefits it has added in terms of you know, the agility and the speed that you can operate now compared to on prem for example.

Yeah, so I mean, we, you know, try to, you know, to automate everything, we try to automate as much as we can, you know, everything from, you know, when code check ins you know, civ pipelines, et cetera. The example you mentioned is fantastic and you know, just really trying to, you know, take the human out of the equation because there's, you know, perhaps better things that they can do and obviously remove the chance of, you know, human error if you're doing an a code implementation. Um so that, that's been fantastic. And I think, you know, we have um you know, different workloads that we run, you know, single a multi a z um and we'll scale up or down the, the compute based on, you know, our peak needs. I mean, it's no secret when tax season is going to be as an example. Um and so, you know, we can be very intelligent around the throughput and, and uh capacity that we have available.

That is actually a very common pattern that we see with our customers where when, when our customers are expecting a surge in the demand, thanksgiving season or tax giving tax preparation season, you can improve or you can increase the performance capability, whether it's throughput or iops right, and scale those parameters down um when, when the demand goes down. So that's, that's i think quite powerful.

Um and you were just, you were also talking about oracle, right? Can you talk a little bit about um what advantages FSx has brought to your oracle applications um in conjunction to the vast unstructured data that, that you're hosting on FSx.

Yeah, I mean, you know, on our on premise, I mean, we run an oracle on, on NetApp quite, quite regularly, same thing with ms sql to be honest, but in, you know, using that in the cloud and having, having FSx for that underlying storage tier really helps out with, you know, the latency, the throughput, the enterprise capabilities that on tap provides. Um and especially around like doing, you know, backups. Um and usually we do use snap mirror, it's been, you know, you don't have to turn the database off to do a backup. Um and that's been very helpful to have, you know, the incremental, you know, backups available, that's, that's pretty, pretty good.

So you've been able to utilize the same storage, not only for your unstructured data needs your c i cd pipelines, but also for these enterprise class databases like sql and oracle.

Definitely. Yeah. Yeah. And, and our approach to snap center is very, I mean, it's the same, I guess because it's, it's the same software. So um we, we use that uh on prem and we're also using that for like the oracle and the ms sql databases in the cloud.

Very good. Um trent. How about in gilead sense is that using any multi a or a single a z or, or architecting the same storage differently for different use cases?

So we are us west one. So unfortunately, there wasn't a multi az available for fsx. Um so we're just doing snap replications to other regions as part of our dr exercise to make sure our data is protected.

And is that in other regions? So you were able to kind of utilize a tool which comes with the storage to improve the availability posture, you know, across two regions. That's correct, right? Without using any other third party tool. That's correct.

So how was that? I mean, i mean, it was, it was easy and straightforward just like how it is on prem right there, there's nothing nuance or new or difficult about it, you set it up and you just let it go. If anything happens to our, you know, site a we just flip a switch and it's running on our secondary regions.

Um talk about what's next for you. Um you know, clearly there are a lot of use cases in life sciences and we've been bringing a lot of innovation, what's what's next for you.

So there's a couple interesting that we're looking to trying with the new announcement that comes up, for example, the simulate access to files and objects to a common files. It could be a potential use case for us for use cases where we use AWS data sync to move files into object. So that's something that we are looking into testing.

Um another is with the scalability of on to FSx on, on a b yesterday. This uh performance as well as the scalability. We are interesting in, in also testing um FSx sends sr storage for hpc and aws in comparison with FSx flusters plus s3.

And, and how do you think, what are the advantages of using FSx for scale out? You know, um is it more performance? Is it larger name space?

Well, for us, I think it's the performance aspect of it, but most importantly, what costs in mind the data duplications and compressions snapshot and also the ability to provide that multiple protocol access. It gives us more options and flexibility to provide access to the storage, not just for the hpc environment, but also potentially for other use cases as well. But you know, we want to make sure that it can do what we think or what, you know the numbers up there it can do by testing it first and it's pretty hard and understand that, you know, with the storage efficiency and the data duplications, does the cost benefit over using luster in s3 makes sense?

So you'll be able to utilize the same storage system that you've been using today for your new higher, you know, performance need requirements and still be able to use all the data services, the data efficiency services and simplify your storage architecture with one single storage option, right? And cost optimizing and that's a big one, right.

Justin about you, are there other things new workloads or new use cases that you're looking at or, or, or is it more of scaling now the environment that you already have?

Yeah, i guess i would say kind of three things. I mean, one is we, you know, we have to complete our um exit of our, you know, last company owned data centers, which we'll, we'll do by the end of next year. Um i think in parallel for the workloads that have been lifted and shifted, i mean, we're circling back around to, you know, modernize and, and, and optimize those workloads as well. Um and then to no surprise, i mean, generative a i is a huge focus. Um for us, we, we've already have two products um launched with generative a i and we're going to be, you know, looking at that quite extensively in the next, you know, 12 months as well.

So, so would i be, would it be fair to say that with FSx, you were able to aol that lift and shift to the cloud and now once you're done with that, you'll be in a position to start modernizing those workloads a lot more quickly compared to you know, another route if you had.

Yes. Yeah. I mean, it, it's, you know, um, i forget exactly. I'll try to mention it but, but when you're doing like a migration, um, i mean, if you start modernizing all at the same time you, it takes longer, um, and you can get yourself, or at least we got, you know, perhaps, um, caught up in like, some complexities and things. So, what we found to be more successful is like, you know, just get it into the cloud. Um and then you have a playground where you can turn back around and modernize and, and because of um you know, fsx being available in aws, that initial migration has been a lot easier.

That's good. Any advice to our customers and partners who are sitting here in terms of as they are thinking about their migration or modernization journey.

So i think it's the technology is not usually the problem, it's the planning, understanding your use cases plan and then cost, right? Understand your cost estimate how much this cost is going to cost to cost you when you move it to a bs, get that evaluations because at the end of the day, you wanna make sure you move once and you don't want to move back, you know, things like macs and pc si um gateway doesn't support mac, right? Then you know, it creates a problem and it also gives your business a sense of you know, maybe don't trust you as much because you, you're disrupting their day to day activities. Right. So, plan and plan and plan and test and then before you actually migrate, that's, that's pretty good advice.

Justin any advice from you.

I agree with all of that. I would also say i like, like verners frugal architect this morning. There's a lot of really good points there. Specifically, i would say, like, hire a finance person that likes tech and, and, and, you know, wrap your arms around them because they're going to be your friend and, you know, just, just don't obsess over the problems too much because maybe they won't happen. So do some good planning, but like, just start getting, going and get some incremental successes and build on them.

Very good.