AWS On Air ft. Amazon S3 Multi-Region Access Points | AWS Events

[Music] hello and welcome welcome peter and welcome welcome ian um you know before we continue before we start talking about the technology before we start talking about the new features let's talk about people who are you introduce yourselves uh peter um you are just kind of here uh peter tell us who you are what do you do and um what brings you here sure i'll look up as i talk to you darko so we kind of maintain uh so my name is pete damick i work on the s3 product management team so my job here is to listen to our customers and then try and convince ian that this is what we should build and why it's the the best thing ever and then work with the end to go deliver that so that's what i do here at aws were you based out of i missed that i'm in boulder colorado boulder colorado wow that's that sounds amazing um ian ian the same questions for you hey folks uh my name is imagary i'm a senior software development manager at amazon s3 i'm based in vancouver bc british columbia in the great white north so all on the american time zones i'm not as far away as you darko but still a bit of a driver a fly up north um i i lead the teams within s3 who have helped to deliver multi-region access points yesterday for storage day um and peter is very effective at convincing me to build things um i will i will say that including multi-region access points so uh yeah super excited to talk to talk to you all about it today awesome awesome well um you know let's let's first of all let's let's ask the basic question what is this thing that you're launching what what is this thing you have to launch what is s3 multi-region access points sure so darko uh jenna for the last 15 years s3 has had our buckets in individual regions and when you wanted to talk to a bucket in a region you had to tell us which bucket you wanted to talk to and which region that bucket was in and our customers have really kind of grown up with s3 and now are developing applications and business tools that need to be able to access data in multiple aws regions and so a multi-region access point its job really is to have a single global endpoint that represents uh right now 1 to 17 different aws regions that you could have data in and have a single global endpoint then to reach that data so so so let me get this straight um let me bring a bucket it's a cup but it's a bucket so i am a customer and i put all of my files in this bucket if all of my customers are sitting in germany they will access this bucket which is launched in germany they access it through germany if i have customers wanted to reach it from japan or from the west coast they also have to come to germany a long way unless i replicate my bucket right is that the case here that's right so you're going to be replicating that data but once you've replicated that data now are you talking to a bucket in germany two buckets yeah two buckets which bucket are you talking to darko okay exactly yeah that's the point right so so in that case the data can be out of sync potentially is that the case that's right and so s3 cross region replication is our technology that allows us to either send that data to multiple buckets if it's all being written to darko's bucket in germany and we want to spread that data out to multiple regions we can have multi-destination replication for that put all of that same data in those multiple buckets but let's say you wanted to synchronize that data so that data written to either bucket would be the same in both locations that's our bi-directional replication with s3 on our cross-region replication okay and and these so these access points basically they give you the ability to automatically detect please correct if i'm wrong so if you have a specific access point and my customer access it accesses my bucket from anywhere in the world they will always be pointed to the closest bucket is that is that the case or am i getting this wrong no you've got that right so actually this is where i'm going to have ian talk a little bit about the secret sauce and magic behind how we accelerate those requests yeah absolutely so the the very cool thing about multi-region access points is that we we took together we took a lot of existing aws technologies including cross-reaching replication and we've added the secret sauces that peter just referenced which is global accelerator and private link to be able to achieve these kinds of cross-regional or optimize routing so you know it's it's in our blog post and i encourage folks to take a look but we talk about the 60 performance increase up to um that's all because of the global accelerator network that we have underneath and that's a product available today to customers but we've leveraged that for multi-region access points and now what you'll find is that wherever clients are based they will be routed to the closest bucket to them so if for example you had a bucket um in dublin you had a bucket in london um and you were based down the south of ireland where i'm from um and you were putting your pictures up there you would get routed to dublin not to london but your data could be replicated across the london so people over there can access it too and vice versa for somebody in london they would get routed to the the london bucket but but the access point is always the same right this this point okay so that's the difference because i recall a long time ago you know s3 replication was a thing but if you wanted a customer to access your specific bucket in a specific region they would use the url for that region right the access point well another access point but they would access a point uh access that bucket through a specific means okay okay got it got it exactly and this thing also replicates data right so uh multi-region replication uh doesn't sorry multi-region access point do not just create access points but they also wrap do the data replication as well well so the way that actually works it's kind of layers of a cake so if you think of kind of how ian talked about these existing s3 features the multi-region access point just sort of acts as sort of the top layer of the cake so it actually sits on top of the existing s3 features including s3 cross region replication so that's actually the technology that's being used to move the data so the multi-region access point his job is not to really know what's in the bucket but to find the the lowest latency copy of that data and what we do with cross-region replication is that's what is actually moving the data and putting the same copy of the data in multiple buckets so that when you reach that global end point which is your multi-region access point when we make that request like gain said to double under london it's the same data in either bucket so we'll just pick the the fastest lowest latency path to that copy regardless if you're in london or dublin or in the us china wherever you might be right awesome so you've talked a little bit about performance here um can you talk like help us understand which customers are gonna benefit from this and why would they choose to use this sure so in this case one of the main advantages that you talked about is that global accelerator network so that's where we have all of those edge locations where we can take let's say an upload request for a photo or let's say it's a large video file or a large chip design a high resolution chip design even some of those can be well over a terabyte and so when we want to get that up into s3 as fast as possible as an upload we don't want to send that across the public internet we don't want to go through multiple hops we want to onboard that onto the aws global network at the closest local point of presence that closest local network point of presence we onboard that upload transfer that again to the lowest latency bucket and that's where you really see the performance advantage of this multi-region access point using the technology that ian talked about the aws global accelerator now a question here is uh we have a question from chad and thank you please do post your questions in the chat as well from from dan lancaster um how does file locking work um same file is opened up or edited from two different buckets or being uploaded with the same name is there some sort of a way you handled those conflicts there is and the beauty of it again kind of using our cake model with cross region replication your buckets that you want to replicate between those have to have versioning enabled so versioning is kind of the foundational feature here that allows us to create multiple versions of the same object as you upload new versions of that same object or go to delete that object we just store each revision history as a specific version of that object so when we go to replicate we replicate those versions as well so that's really how we handle any sort of conflicts is if you're writing to a bucket and you want to replicate that bucket versioning has to be enabled first and that's what allows us to keep track and then sync the versions from both sides so this feature really kind of gets you closer to an active active s3 configuration where you could upload videos from your your cam to any bucket in the world and then we're going to sync that data with that cross region replication technology which is built again layered on top of s3 object versioning where we store each of those objects independently of the last one awesome awesome i love that we have another uh we have another question from chad that i think is really interesting um this one's from santosh how does data privacy regulation compliance work with multi-region replication i love this question uh this is actually something that's different about multi-region access points than some of the other solutions that are out there that are similar but obviously a little bit different so let's talk about the differences here with cross region replication being the engine of the data movement that's actually what's transferring the data between the buckets the cross-region replication rules themselves are very precise they let you control what data you want to replicate to which buckets so you do not have to have the same data in all of your buckets if you don't want to you don't have to replicate all of the data in your bucket you can select just at a prefix or a folder level or you can even select at a tag level and those are called scopes inside of your replication rule so jenna what's the scope of replication that you want to to replicate tell us the what is it just a folder is it everything in the bucket or is it just maybe data that's tagged with this tag so if there's data you don't want to go across geopolitical borders that simply would then be outside of your replication scope you wouldn't include that in the replication scope behind this multi-region access point and do you also choose which regions to replicate to and from correct so you have complete control it's very precise okay and lets you have that control of what data goes to which regions and is in which buckets uh so that is one of the advantages of of what ian's team has built i think just the fact that you can choose which part of your bucket will be replicated is just chef's case i love that fact because gdpr data regulations all of those things are quite important and we tend to funnel a lot of data in a single place data lakes and all that stuff so i think this is this is really really good for that so i love that i love that thanks and the cost is also a big factor some customers simply have so much data that it's just not feasible for them to replicate everything right so absolutely that is one of the big differences in the solution versus others is that that precision of what data goes where right right right um there's a question from somebody here on chat as well i think i missed it from uh yeah so from okagan 2000 now this question may not make sense because how does uh how the global endpoints find objects if it's missing in the bucket for example if i go to a bucket in my re my local region can i see those different buckets can i see those individual buckets which this is being replicated to or this is like a global thing like uh sure if i delete the file for example in a if i go to a bucket in ireland and i delete the bucket delete a file in ireland does that does that file get deleted all over the place or that's a great question that goes back to our precision of our replication rules so on the rule itself we can actually instruct the replication engine to replicate deletions or not so if you want to have your deletions replicated and propagated to your other replicas that's simply a single checkbox that is completely up to you so some customers want that safety and security of i don't want to replicate my deletions i want to just keep the data in each bucket kind of unique i i or i do want to replicate the deletion so i kind of keep them in sync i want kind of a mirrored approach where i want to replicate deletions and i want to sync the data and the metadata and those are specific options in the replication rule and that also includes an option called replication time control that's our premium replication offering that allows a customer to say jenna i want a sla with financial credits back to me if sfaws can't match this sla can't meet this sla of 99.99 so four nines uh of all of my objects 99.99 of all of my objects have to complete replication within 15 minutes or less and ian i don't think we've paid out on that sla yet we haven't missed it yet so i think our track record is pretty good when it comes to replication time control and that that sla so if you want those buckets in sync and you need an sla on your replication that's really where replication time control comes in again it's a single check box in your rule for replication that's a good point how long does it take to replicate right so if if if i make a change in in ireland how long do i have to wait for that change to occur in tokyo yeah most objects complete replication in seconds to minutes but with replication time control that's where you get sort of that finite end fixed post of of an sla of 15 minutes or less for 99.99 of all of your objects and we haven't missed that sla yet okay amazing i'd love to hear that one thing quick thing i want to jump in on as well because we spoke a bit about the the edge locations in the internet use cases and i do not want people for to forget that we've also launched with support for private link and so you can privately access your data from within your vpc you set up a vpc endpoint specific for multi-region access points and then you can access your multi-region access points over that and the very cool thing about that is that extends multi-region access points to on-prem if you're using direct connect or aws vpn and you can set up a multi-region access point to cover different regions um some of the us regions and then you can have your vpc in dublin and you can be privately accessing that data replicated across those u.s regions if at all from your vpc in dublin so it's the private aspect of it is is definitely a very very cool aspect that i think folks will love that was actually one of my next questions like like private vps endpoints that is important so folks out there vps endpoints enable you to access specific database services not through the power of the internet but through the database backbone so without actually ever leaving your database network so uh this feature is also very important if you want to have this uh set up as well so amazing amazing i i like that thing so good good good good also just just before we continue i want to welcome all the new folks in the chat welcome this is aws on air this is a weekly show where we get to talk to our wonderful ghosts ghosts [Laughter] our wonderful guests guests and hosts guests and hosts a wonderful wonderful guest from the aws service teams and talk about the newest newest releases today we have ian and peter here talking about the latest launch from amazon s3 called multi-region replicate multi-region endpoints um so yeah feel free to ask us questions put the questions in the chat and we would love to have them answered right now okay um i have a question i have a question how many regions is this available in yeah this this one is uh available in 17 regions at launch okay um so it's quite pervasive across aws which is really really nice i think we'll find that will make people's kind of brains start thinking about wow well now where where can i put my data and where can i now reach out to my customers right and make their lives better um we've talked spoke to a number of customers about who are now i guess international right and their their software crosses boundaries and so now they can think about improving the performance that their own customers will see when using their software and regarding encryption there's a question from a linkedin user unfortunately we cannot see everybody's name but that's linkedin so linkedin user asks does this support the encryption of specific data addressed and data in motion yes so there's three different types of of s3 encryption at aws and so uh the multi-region access points they are encryption agnostic so they're not aware of the encryption status of the data but the replication engine is so with sse-s3 replication that's our server side encryption that is supported with replication as well as kms keys on those objects can be replicated the only one that is not supported today is customer provided keys so the sse sse-c so if those objects are encrypted with sse-c those cannot be replicated today with our replication engine but that is something that we're looking at in the future and and that's regardless if it's behind a multi-region access point or not that's just a limitation of our current replication engine okay yeah good point good night so if you're using your own encryption keys in an s3 bucket this is not working yet for you so that's correct stay tuned yeah and just to cover the data in motion part i mean you we also support tls encryptions as well so okay there's no um there's no concerns about data in motion being plain text or anything okay good good very important right yes that's the auction part yeah yeah yeah so okay good yeah so it's also being encrypted in transit as well so good awesome awesome um now kind of a question here what kind of users do you see kind of really benefit benefiting from this i know peter mentioned uh the chip manufacturers like those high resolution chip images for example but what other customers what kind of other use cases do you see customers using the service for i can yeah i can jump in peter has has many but one that i think is close to home for myself um that i know about is just being able to provide customers with a single endpoint that all their clients can use to upload data optimally so i kind of mentioned that a bit before about having you know globally dispersed customers or clients that are trying to get data often so think of mobile phones think of tablets laptops these kinds of things they're trying to get data up to the cloud as quickly as possible um the way customers have solved this in the past is they have complex client logic baked into the client that will choose the right region based on where they are or they'll have a proxy layer in between that will take the data and then decide where to go from there now you can just get rid of all of that give it a single end point and let it aws decide um where to route it for the best experience so i think that's just a really nice use case and i'm excited to see that one be taken up by customers excellent awesome awesome um and then we we we make a slight change of our uh asl interpreters thank you very much um cool cool um so i have a question i got another question what so i have some s3 buckets out there right now what do i need to do if i want to start using this what is what do i do how can i can i start using it with my existing s3 bucket sure we can actually jump in and show you what that looks like here but it's super easy jenna uh ian's team has done a great job you can use your existing buckets that have existing replication rules on them or you can make new buckets or you can mix and match existing buckets and new buckets so there is no magic bucket that you have to create or a special type of multi-region access point bucket these are just existing buckets that you already have that you want to have a new global endpoint in front of or these can be new buckets that you want to start from from scratch with and so it's completely your choice okay and then what do i have to make changes in my code or like what what's the extent of my my code changes that i have to do yeah ian can you talk a little bit about our s3-global yeah i mean this is kind of the interesting thing that we we touched on at the very start about well before i had to configure a region um now it's not necessarily a region more so that you're configuring a global right so it it is a global region or global name space is the best way to think about it and so your and your endpoint that you use at the very end doesn't actually have a region name in it it has global in it um so s3-global is now our new i guess global name space in that sense um and the the work or the extent that they work to update this to start using it is obviously the configuration yourself in the console or maybe sdk cli or cloud formation um to get things up and running choose your buckets create your multi-region access point and then to update your client to now point to your multi-region access point instead of pointing to your pocket excellent okay got it so then the name will change jenna so this will have kind of an interesting looking endpoint name to it and that's because obviously security is job one here at aws so we obfuscate that endpoint name we generate that for you so it's not gonna identify your company in it uh you can't choose the name itself but the reason why we do that is obviously that global endpoint can be representing a lot of buckets and a lot of data in a lot of different regions and so when you go to talk to that end point you're no longer talking to a bucket so the the format does change as well you're actually talking to what we refer to as an arm or an amazon resource name and so that resource name is going to have a slightly different format to it but you'll simply use that that format for your your gets your puts your uploads your downloads your list requests those data operations you'll just change sort of the the the pattern or the end point really that you're talking to here and what what ian did a great job with is in the past you've had to specify darko would have to specify us-east-one or two or you as this is the location of the bucket that i want to talk to but what's new in this new endpoint name this global name is the region gets replaced with like a wild card called s3 dash global that s3 dash global is the new global region for all of the buckets behind it so that's kind of how these two pieces fit together so the change on your side as an application or as a client or a developer when you want to interact with s3 is you will use a different endpoint name can you still access the data in the buckets directly like you always have absolutely it's just you now get a new global endpoint name that represents sort of like a bucket group exactly and here's an example uh i've i've i've ran into this issue so if you ever needed to install the for example the code deploy agent on your ec2 instances or even the ecs agent if you go to our documentation and look to download one of those agents they will usually direct you to a bucket nearest to you by actually entering the url of the bucket nearest to you so that thing is still that well it's you know this is a new service so but if imagine if that was applying multi-regional access points there would be a single url and it would be automatically take you to the bucket that's closest to you which is just a wonderful thing that's right so you asked like what type of customers would be interested in this ian gave a great example media companies with a lot of video and photos that need to be uh uploaded but the other one is kind of one you just mentioned darko which is kind of like if you're a software company for example and you have to make the daily build of your software uh package available to your engineers around the world well now you can replicate all of that data with s3 and still give everyone a single global endpoint to go to get that that daily build from to pull that down and not have to worry about well which bucket is closest to me and which bucket should we talk to and if we have you know a connectivity challenge to that bucket on a tuesday well now we have to go look for the next bucket you don't have to worry about that anymore you just always now have that single global endpoint to talk to yeah exactly exactly excellent excellent uh so that's very good especially if you're like looking to get data faster to your end users in specific regions now that data will be automatically closer to them and they can automatically download it from the same url which is just again wonderful um this brings always a discussion look we talked about actually buckets and global presence and and global distribution the first thing that comes to mind is content delivery right content delivery network aka cloudfront um there's an actual there's even a question here from from somebody in chat i think i missed it but uh okay i'll get to that question if i find it i apologize there we go thank you thank you thank you linkedin user um is there any attention points is there something we need to pay attention to for s3 multi-region access points over cloudfront so if you have a call if you're using s3 as a cloudfront distribution does this impact it in any way it does today you have to make a choice so cloudfront is if you're up and running with cloudfront today uh fantastic keep using it but you cannot use a multi-region access point as a cloudfront distribution origin so those two are currently separate things but if that's something that is appealing to you as a customer uh ian and i are here and ready and waiting for for your feedback if that's what you'd like us to go build next uh we're we're interested in hearing more about that but so today darko to answer your question those are separate uh they do not today uh integrate with each other good to know i think it's very important and you know folks chat twitch linkedin wherever you're watching us from if you have ideas for a feedback service request let us know right it can be through us or it can be through through your normal channels we want to make the product that's the best for you so uh you know uh just tell us what you need so uh awesome i think we're are we ready for a demo yeah i think i think it's the demo time um peter uh we were we were told that you were gonna show us something um so uh when you when you can't share your screen i thought it was ian it was me you know i'm just kidding i'm just kidding [Laughter] thank you pam okay okay peter let's uh let's bring that up um while you're doing that maybe maybe maybe maybe you can actually answer the question from the smallest one uh i am policies does that all get replicated bucket policies all of those things uh is that being replicated across the board as well so when you create your multi-region access point um you have the ability to create a policy with that so it's an actual multivision access point policy and then that gets replicated to all the various regions so that will apply and it's the same thing with your vpc endpoint policy um if you create a vpc endpoint policy that also gets brought around to all the regions with with and with multi-region access point private link so those go around so it's a separate thing you create for for your uh for your tonight okay excellent thank you um okay so um alex marsha is mentioning the demo gods pray first so let's see uh if peter has submitted all the things he needs um okay peter can you make that bigger like a lot bigger i can try just just control and plus and all the stuff that you need so got a lot of white space here yeah can you just control and plus is that a possibility yeah or see if we can just make this full screen here no no css right um [Laughter] so so i mean if you can just zoom the content in on the screen right so um if you want just in the browser just zoom in the browser that's fine you can you can you can maximize it just zoom in the browser and that should be fine any better perfect yeah perfect maybe just a little bit more uh that's good let's go excellent yay all right so when we go to create our multi-region access point there's not a lot to it really we're looking at just a couple of data points here so we need to give it a friendly name now this name doesn't really have any bearing it's just this is going to be what you're going to refer to this as internally kind of like a label if you will so we'll give it our twitch tv name and now here's where we'll go and decide which regions and which buckets we actually want to include behind this multi-region access point we can create a new bucket but i think most people were interested in selecting an existing bucket that they already had in place so with our multi-region access point you can select again one bucket per region up to 17 regions if we try and select two buckets with the same region it's going to come back and tell us too many buckets for that region and you'll see here a little warning telling you sorry you've selected two from the same range okay so when we select a source and a destination bucket here we'll take a data lake example we'll take a destination bucket here one in west two and one in east two here in the u.s we'll add those buckets not a lot to it so now we can see we've got our two buckets selected here and now we have to choose for the endpoint that we're creating this new global endpoint how public do we want to make this endpoint by default all of our buckets are private and our multi-region access points are private as well so we enable block public access by default we create this new endpoint and that is literally the only two steps that you have to choose to configure or to create this new multi-region access point i don't think we could have made it any easier but i think this is pretty good for such a powerful feature really it's about two maybe three clicks at the most to create this now while we're doing this this is actually asynchronously going out and setting up all of our certificate infrastructure at each of these regions so ian do you want to talk a little bit about what's going on kind of behind the scenes here from the infrastructure side inside of aws yeah absolutely so while this is basically creating we're setting up global accelerator we're setting up the dns entries provisioning the certificates the tls certificates you'll need for tls encryption um and configuring all of those to go across to these two buckets in the regions that pete has selected um and so if you think about when you're going to 17 if you create a multi-region access point that has 17 regions behind it we're going to positioning all those resources and infrastructure behind that as well but it is quite fast which is a really really cool thing about it [Music] and uh is is this one ready pete so i i just showed one of the examples that we had with 17 buckets all behind this one single multi-region access point just to kind of show what it looks like here and so so so this is an example of how it works i need to i need to crouch underneath this thing this is an example where you can see this this is twitch one second hi uh so uh this is an example of how how this works when you create uh when you use an existing bucket right uh but if you create a new multi can you create a multi-region bucket out of the box so it will create the buckets for you in other regions or do you have to already have those buckets existing somewhere yeah so when we look at our multi-region access point uh creation flow let's go back and take a look at that and create a second one here so we go to create a multi-region access point we'll jump back into that we'll call it two now here's where we actually have that option so if we wanted to create new buckets uh in the region that we wanted to use it's just going to then take you to the normal existing s3 create bucket flow so there's nothing special or magical about this in terms of of your buckets you don't need to create new buckets but you certainly can and it's just going to take you to the create bucket flow right here and if you want to use existing buckets well this will be where you'll add those existing buckets now you might ask also which is a common question well can a bucket be in more than one multi-region access point and the answer is absolutely yes so i could take those same two buckets that i just selected my data lake and my destination bucket in west 2 and east 2 add those to my new multi-region access point and i can go and create that new endpoint as well so there's no sorry i guess i need to have a better name here right call it twitch2 now when we go to create our new multi-region access point even though it's the same buckets that we just selected for our first multi-region access point uh that we now just have two separate endpoints that can either one be used to access the data that are in those two buckets i love this excellent i i i love it when it's simple so uh excellent excellent so um all right um let me just bring in so let's our twitch tv one is ready uh he is done and so now there's really not a lot of additional configuration to do once our twitch tv multi-region access point is complete here's our two buckets the only really two options to set up now are the permissions which is going to be the the policy that ian talked about earlier where we will set up a standard imes language policy for this particular endpoint you can see our block public access is still enabled here and then the second option is completely optional this is where we would set up the replication rules now could we go to the buckets directly jenna and d'arco to set up replication absolutely there's no difference in the replication being set up here with the multi-region access point or going to the buckets directly it's the same engine same replication rules same options so you'll be able to create those replication rules right from here and then we'll actually even give you a replication metrics view if you've enabled cloudwatch monitoring on your replication will tell you how much latency is on that replication rule will also tell you how many objects are lagging behind and will also tell you how much data in terms of bytes the replication is lagging behind you also get a nice replication summary here that customers really have loved ian you want to talk a little bit about this replication map feature yeah absolutely i think this is one of the coolest features it it's simple i can't get complicated when it's it's up to 17 buckets but if as you can see you're basically selecting for this your two buckets and you've got arrows showing that you've got two-way replication turned on for these buckets between these regions um if if it's one-way replication you'll also see that and then you can basically scroll around the map see where your data is replicating and it's a nice way to visualize how your data is moving around the globe which um is not easy to just read from a table i think it's cool to see it on an actual map is there an example of the 17 buckets can we see that i do do you want to see that it looks a little crazy so it's a little a little uh but yeah i can jump in it's a friday evening we can go crazy that's okay i think my manager is listening to this as well so uh hopefully i still have employed here monday uh but i just wanted to show you what a replication rule looks like here this is how easy we've made it it's basically kind of templatized so if you want to send out data from one bucket to a bunch of different destinations that would be the choice on the left if you want to have that bi-directional sync replication where we're synchronizing the bucket contents that would be the choice on the right and now we can choose a little a little bit of our replication options we want to enable the rule and here is jenna's question about how how broad of a replication scope do we want do we want all of the objects in the bucket or do we want just a prefix or a tag our replication is going to include any encrypted objects and we're going to encrypt those with sse-s3 and then finally here are options this is where we decide if we want to synchronize the buckets bidirectionally this is where we decide if we want to replicate deletions this is where we enable the cloud watch monitoring for the metrics how far behind are we how much data is left to replicate uh how many objects are left to replicate and then finally kind of the the cherry on top replication time control this is again our premium replication option where if you needed that sla 15 minutes or less for 99.99 of your objects this is the option for you and in this case the the um the monitoring and the metrics are automatically enabled when you choose replication time control i love talking about it's one of my favorite features to talk about uh it's just pretty unique uh with that sla on top of the replication wonderful any questions about these yeah no this seems pretty clear simple great okay so now you're ready for the 17 uh buck it's a little spot let's take a look here let's see if your laptop survives um okay it takes a moment to load i guess well while we're waiting for that to load are there um any customers that are already using this that that we can talk about that's a good question i don't know if i can say their names i can tell you what industries they're in uh we have some automotive customers we have some iot customers that want to send us telemetry we have some analytics customers on the daily side and we have some customers on the financial data analysis and subscription business that are so what you're saying so what you're saying is it's used for multiple industries right it is everybody jenna needs faster s3 yeah exactly uh you know that that upload and download it and in this all right and in essence this is still an s3 bucket underneath so the same all the same features and benefits storage tiers glacier deep glacier all of this is part of this even though it has a multi-access endpoint or amber app as i'm going to call it not because i've seen the acronym uh it has the same things right it that that's right and that's that's a great point on it in that you can if you are today already using cross-region replication to move your data around and then having your different clients access that data either through configuration where they go to separate endpoints or a proxy layer or some baked in logic you can take it out you can put in a single endpoint and and let um let everything else be taken care of so i think that's a really nice aspect in that you don't you you can take existing workloads and put this use this you don't have to uh create everything brand new i think we lost peter so a couple of things the demo just crashed his laptop or he got fired he didn't get fired i think it's i think it's something i think peter had issues with uh technical issues so yeah folks this is live so yeah peter i i hope peter will come back here but um yeah speaking of getting fired um how much does this cost uh we need to talk about pricing right and and those things can uh well we i think we need to get peter for that peter is the right person to talk to us about that okay okay well um maybe maybe maybe ian you would know who is the why wouldn't customers use this thing right i think it sounds wonderful i think it's great i think it's just a nice nice way to kind of get your data across the world and have a single access point for that data um why wouldn't you use this i think um i think the key here is data mobility right and then a geographically dispersed set of clients or customers that's really where this comes into account if you are using s3 in a single region you're in ec2 you're accessing your data and that is working that's really and that's that is i think you know a vast majority of customers of s3 are using it in that way then you mrap is not something you will need at any point in the future but if you start thinking hey i need my data in different locations because of um compliance or i need my data in different locations because of my my customer base is growing internationally that's when you want to start thinking about it so it it may not be right for you right now but it also it may in the um close future if you're looking to expand so i think that's um probably the best peter if if you want to jump in on that as well yeah and welcome back um it was where is mrap or where is multivision access points not a fit if you only have a single bucket if you really are just in need of a single bucket you can use s3 transfer acceleration today with that bucket and then use cloudfront if that really is you know just a single bucket is all you need i don't think there's any reason for a multi-region access point in that use case um i i don't think it's a good fit for that s3 transfer acceleration and multi-region access points have similar performance benefits on an upload so i think that would be one of the areas where i don't think it makes a lot of sense to to have a multi-region access point awesome and peter we were waiting for you we want to talk about pricing you know i understand that sometimes pricing can be a complex thing but how does pricing work on on amraps and multi-region access points with s3 sure so there are two charges associated with a multi-region access point and we tried to make it very much usage based which is kind of our standard here at aws and what i mean by that is the endpoint itself is really where we have the meter running so the meter is based upon how much data you are routing through this endpoint so this new multi-region endpoint that we've created here again uh hopefully you can see it darko here's our 17 regions behind a single multi-region access point so there we go every gigabyte of data that you send through this new endpoint there's a charge associated with that in addition if you are using internet-based uploads through this multi-region access point and want us to accelerate that up to that 60 number that we talked about there's an additional charge for internet-based acceleration for uploads and downloads where we take that traffic off of the public internet bring it onto the aws global network that's an additional charge and it's based upon the length of the hall of the the transmission so if it's coming from north america to australia to sydney for example that's a different charge than if it's all in the same continent of north america u.s east 1 to u.s west 2 or something so that is the second charge based upon the internet-based acceleration that we're doing and the length of the traffic that we're sending across our our global network gotcha okay excellent make sense and thank you this this looks wonderful i actually just took a screenshot of it and you can hover over each each site to kind of get an idea of what's going on uh at each of the the different regions i should say how many replication rules are configured for that bucket in that region so you can kind of get an understanding of of what's there and kind of drill down excellent well um peter ian thank you very much um we want to encourage the audience to check out the blog so we're going to be posting the link to the blog post in the chat here so i think i can do it now on twitch but i need to figure out how to do it on linkedin as well so uh make sure to check out the blog post on this launch and yeah if you if you have a an s3 bucket that's sitting across the globe and you want your audience to be able to access that bucket from anywhere from a single endpoint check out uh aws um mrap or abs or sorry amazon s3 mrap multi-region access points for amazon s3 right perfect nailed it there we go uh excellent uh peter ian thank you very much uh it's been a pleasure to have you here uh and i'm looking forward to actually trying this thing out so thank you [Music] you

Channel: AWS Events

Views: 705

Keywords: AWS, Events, Webinars, Amazon Web Services, AWS Cloud, Amazon Cloud, AWS re:Invent, AWS Summit, AWS re:Inforce, AWSome Day Online, aws tutorial, aws demo, aws webinar, #tutorial, #demo, #s3, #awsonair

Id: xRMfItnVrq8

Length: 47min 28sec (2848 seconds)

Published: Thu Sep 23 2021