Hello everyone. Uh I'm Rishi Karaman. I'm a product manager for Google Cloud Net App Volumes. Um this presentation was done at the mini session in uh Google Cloud Next. Uh it was co-presented with Danny Vas who's a product manager from Google. Before um we get into flex um quick recap of Google cloud NetApp volumes uh it is a fully managed uh storage service by NetApp um provides multi-protocol for enterprise workloads u at a native experience including full integration to Google IM API and cloud console data protection and data management is built in uh this comes from NetApp so it's NetApp is known for its uh data protection and scalability while providing snapshots, volume uh clones, uh replication and backup integrating into Google. This is secure by design. Uh network security and data encryption at rest are standard. Um in addition to providing you full access of your encryption with customer manage keys.So let's look at what uh what we've been offering in terms of Google uh service levels. Google Cloud App Volumes has three service levels right now. Standard, premium, and extreme. Each of these service levels is determined by the throughput it provides per terabyte of capacity provisioned. Um, standard starts at 16 Mbps. You can jump up to premium which is 64 and extreme at 128. Across these service levels um some common themes are the minimum storage pool limit which is sort of two tib the minimum volume limit it's 100 gig and all three service levels are characterized by dedicated per volume performance um which means thethroughput actually depends on the capacity of the volume provisioned not just the storage pool. So what our customers have been saying is um in addition to getting all the benefits of the three service levels, they would want to see a capability to provision lower volumes, right? At the same time, just because the volumes are smaller, they don't want the performance to be throttled at those lower limits. Um and in addition, they would want to see more resiliency in terms of explicit zone failure protection. So flex is the sniff service level that we're introducing which comes with um some of the flexibility to provision smaller pools and volumes. We can go as low as one tab and one gabyte volumes. Uh all the volumes in a storage pool share the performance. So the total performance by the allocated size of the storage pool is available for any volume in that pool irrespective of the size of the volume. And then it offers two flavors. One is a zonal availability where you select a single zone and your volume is provisioned and is localized to that zone. Whereas with the regional you have an option to choose another replica zone where your data is synchronously replicated. We will deep dive into that um into that soon. Before we get into those salient features that are new, let's just look at what it shares with um the existing service levels. Flex will be an additional service level. So in regions where you have standard, premium and extreme, you will now see flex and flex is actually built on top. So it provides you the same level of um protocol, multi-protocol and uh data resiliency in terms of having NFSV3, V4.1 and SMB and all the enterprise man data management capabilities like snapshots, clones, cross region replication and so forth at the same time carrying over the security elements of data at rest encryption uh and the full integration withGoogle cloud. Uh Flex will start out at um at in the 15 regions. One element of Flex is that it'sbuilt natively on Google hardware. So it is um it is very agile in terms of deployment. It will be available across all GCP regions um very soon beyond the 15. We talked about the flexibility in terms of capacity and performance sizing. So we start out at one temp and 100 one gig volumes. This allows you to have really small volumes especially for your DevOps and CI/CD pipelines where you require a lot of small volumes Kubernetes workloads and so forth but at the same time it allows you the ability to burst performance in certain periods. And the third piece we already talked about this it allows you the zonal and regional protection. So which means in case there was a failure of a single zone where your primary volume happens to reside um the data would automatically be failed over to the secondary zone and you would have continuous access to your volumes and the data. Double clicking on the shared performance. So in this example you see that there are different capacities of volumes provisioned. uh the first volume which happens to be the largest volume actually is not using most of the throughput in this scenario. The total storage pool is capable of delivering 1 Gbps and as you see here 10 GB volume and even some smaller in some cases is capable of driving almost 90% or um to the full capacity of the storage pools provisioned storage provisioned throughput.Um and again thisdynamically scales. So at any given point of time one volume could be using most of it and um while that volume is not bursty some other volume cantake on. So of course this means that it gives you the flexibility to uh have different workloads active at different times. The 15 regions that we talked about where this will be available are listed here. It's available across the globe. Um but it's a matter of expanding it um pretty quickly. Our plan is to be in all GCP regions by third end of third quarter of this calendar year. All right. So it's time for a quick demo. So we're going to go ahead and kick off by creating a storage pool. This case we provide the storage pool name. And name. And the first thing you do is select the region where flex is available. one of those 15 and soon any region. Select your service level. In this case, we select flex. As you see, it's allowing you to go as low as one tip storage pool. Uh the next step would be to select your availability. If you select zonal, you get to choose a zone in which the pool will be deployed. By selecting regional, you have the ability to select region replica zone as well to which the data is synchronously replicated providing zero RPO and business continuity.We're going to yeah so it is across those two zones. Um we're going to provision a one tip volume which is the smallest that one tip pool which is the smallest that's allowed. Um you select or VPC and create the storage pool. That's the first step. Once you have your Plex storage pool ready you can see theprimary zone and the replica zone. Um and that will replic that will indicate where your volume reside. So then next step let's go ahead and create a volume in one of those pools. Um similarly you provide your volume name. You select your storage pool and capacity. So you can go as low as 1 gig volume. So in this example you're going to create a one gig of volume and know that this one gig volume is capable of driving the one tip throughut or the one tip capacity based throughput. So, we're going to mount this volume based on the mount point shared on the in your screen. Um, we're going to run a very simple fio command just to do um a read ofyour from your volume. And this will soon jump up to the full 16 Mbps that is allocated to that storage pool uh running out of a one gig volume. So, that showcases howthe volume can burst up to the complete throughput available for that storage pool. As your storage pool increases, the 16 Mbps gets multiplied. We're going to now simulate a zone failure bybringing down some nodes in zone A, which is where, if you remember, the primary zone was selected. Uh you see a short uh pause in the IO up to 2 minutes. Um that's the time by which the um storage pool fails over automatically to the replica zone. And once that failover is complete, you'll see that the IO resumes um immediately. You don't need to do any operations on your end in terms of remount. It just resumes from the other side. You also notice that um the zones have switched. This will also reflect in the storage pool. Every volume in the storage pool automatically switches over to the secondary zone. That's the regional availability of um Flex. So Flex in addition to all the other cool things we talked about provides you this flexibility to choose your availability as well. Quickly summarizing the takeaways. Um, with Flex, you now have the capability to provision smaller volumes to meet your applications without compromising on performance. You meet your availability requirements with either zonal or regional and you have cross region replication in case of uh you want to protect against a regional failure. Now, Flex is available in all regions and zones. um with uh leveraging the Google Google's hardware innovation. It is available in every region and zone you run your applications. So that's it. We are very excited to bring Flex uh to you. It will be available generally available by uh end of May um in 15 regions and soon expanded to every Google cloud region. Thank you.