What's new in Cloud FinOps?

WNiCF - March News 2023

April 28, 2023 Stephen Old and Frank Contrepois Season 3 Episode 15
What's new in Cloud FinOps?
WNiCF - March News 2023
Show Notes Transcript

Frank and Steve dive into what's new in March 2023, with some cool stuff in AWS and Azure and even digging deep into the release notes of google to find some more cool things.

The episode is topped off by learning about some of Frank's really cool research, and for our native French listeners, look out for Frank's new French version of the news!

March 2023


Data/DBs

Visibility (Billing conductor - Tags - cost categories)

Commitment


Savings 

Storage

Instances


Sustainability

  • Nothing 

Misc/Silly


computerweekly.com/feature/The-race-to-net-zero-Six-ways-to-slash-IT-infrastructure-emissions   

https://www.computerweekly.com/news/365531217/Governments-revised-net-zero-strategy-could-put-public-sector-under-more-pressure-to-move-to-cloud




Full Transcript

Hello everyone and welcome to our latest episode of What's New in Cloud Finops. This is the episode for the March news of 2023. Myself, Steven Old, and my friend Hello Frank. Hello. Not loads of news today, so hopefully quite a quick episode. But you've been doing some really cool stuff in the background, which we're going to talk about at the end, which I'm excited about. And I think it's straight to you for some news, yes. So the first news is for me, it's Amazon Neptune. Serverless now scales down to one NCU to save costs. So the Neptune, Amazon Neptune is a database. I think it's a graph database. So it's a very specific niche database. And there is a serverless option, and until now you could only scale down to what is 2.5 Neptune capacity units, but now they've reduced it to one, which means that the smallest instance you can run is 1 Ncu. Which is very cool because if you are just needing, as, most people probably will be trying Neptune just at the beginning, they want it to be as small as possible to keep as cheap as possible. So that's good news. Fantastic. And then straight on to me for the next piece of news. This is that AWS database migration service Fleet Advisor now supports target recommendations, so this has become generally available. And I mean, the clue's in the title, right, But it actually gives advice on what you should be moving to based on a series of metrics. And it's all built into the DMS product, which is in, I say, most regions. 

It's probably not. Yeah, I say it's been around for a long time, hasn't it, as DMS and they've added this piece straight on. So I'm just having to look through the regions and it's in the ones you'd expect. It's in a good mix. So that's. Nice news and hopefully that will reduce the amount of right sizing that's needed post migration. And the problem is, and I don't know if you see this, I see this a lot. People are less likely to right size RDS. Oh yes oh database in general because it's you've already just had the downtime to move and probably been shouted up for your customers. Any horizontal scaling requires you to close down and open. So yeah, getting that right first time or getting that closer to right should reduce that cost. And there's a lot of talk at the moment in the market about people not realizing the value of cloud from a lift and shift. And this is one of the many reasons why exactly And there is also, if you come like me from the DBA world is every time you change one parameter, there are another 20 you need to change in the database engine if you're doing your job correctly. So you end up and say, oh we've reduced the memories, oh crap, I need to change all the settings. So that's by the way. It's an interesting point I always find interesting is that I don't think that RDS TBS are depending on the size the instance that the software will be configured appropriately. Like if you have a Postgres X large, should it be configured differently than a Postgres to X large? I'm not sure you get that. I think you get the standard option. I need to check. That's anyway that's for me. Good point. My next news is still on RDS, it's for Maria DB and it is now supports RDS optimized rights. And we've talked about it in the past is that there is a new way that AWS uses to write on the storage for DB and that allows to be twice time faster. It's not really faster but you can do the double amount of transactions. 

Yeah, so it's really cool. It doesn't cost anything more. And it also helps a lot with logs so that it's guaranteed that the logs will not be corrupted. There are lots of interesting things coming out of this capability because lots of database survive. Really. Thanks to the storage or or as efficient as the storage is fast. Absolutely right. 

The next one for me, and this one's a little bit of a cheeky one. And may I say also how much better life is now. We've found a better way of finding Azure news. Still haven't found a good way of finding GSP news listeners I had this week for you. I went through all of the release notes for March to try to find stuff. So if anyone at Google is listening and you have a better resource for this, please save me some time because that was. Painful. Anyway, this one is from our friend at Microsoft, and it's written by Peter Carlin, the corporate vice president of Azure Databases Services. And why this one's a bit cheeky is very rarely do we find a blog or an article that literally calls out a competitor by name. But the title of this article is Microsoft Store Innovation Powers Leading Price Performance for SQL Server. And when you go into it, it's a report done by GigaOM, the analysts there no the analyst firm and the analysts out there on this firm. And basically what it says is that SQL Server on Azure Virtual Machines. So this is not the managed service SQL Server on Azure virtual machines up to 57% faster and costing up to 54% less than AWS EC2 on a price performance basis with Azure hybrid benefit and the three-year. A commitment. So that means a load of caveats. But very rarely do we see people really call that out. And I thought that was an interesting one and we do see and Frank attests this, the 

importance of kind of following the data and how powerful the data is and the gravity that creates and database is a big part of that as well. And when you are a heavy SQL house and you're looking at this, it's these kind of pieces that do. Grab the attention of those, looking at the Rois and the Ttos and stuff and it's quite interesting. They are right. But quite often it's not really used cuz you never know how much data you're gonna generate up front. You just know it's gonna grow, but you don't know how much data you're gonna generate up front. And it's compounding the interesting piece. But yes, I think we need some sound effect. We need to do the sound effect. I didn't do it at the beginning. 

Here we are in visibility that's Billing, Conductor, Tax, Cost categories, all these kind of things. And the first one is from AWS allow listing tool for testing new billing cost management and account console permission permissions. The interesting bit is that. They they are. There is more new permissions, IM permissions, more fine grain for billing, cost management and account services. Which is very welcome because in the past it was an all or nothing approach. Now it's it is much better. But yet every time you change something and it become more locked in, it's very hard to find the right combination to show what you want. To show you what you want to enable your person to see only what they need to see, especially if you are using the list. What is the? The minimum level of permissions, because until now probably people were just forbidden or allowed. Now you need to go more granular and so they've created a tool, an approach that allows you to test or to generate those permissions in a simpler and better way. That's how I understand it at all, at least fantastic. Still in visibility. And this one kind of sneaks in because it's around. Ads Cost Anomaly detection and Cost Explorer. But basically the news literally at the end of March is that Ads cost anomaly detection now automatically configured for all new Cost Explorer users. While previously you would have to any new user would have to set up enable Cost Explorer, then set up Cost Monitors and then Alerting preferences and then subscribe to an e-mail. The new automatic configuration removes that and with launch. An AWS service monitor and a daily e-mail subscription will be created for all new Cloud Cost Explorer users that are enabled after March the 27th. I mean cost anomaly detection is you know can set up its red homes but for no additional effort. Now people are just getting that bit more 

visibility, bit more viewpoint. If the actual spend is over $100 and exceeds of expected spend, I think then you get the alert to standard. So it's a very standardized rule, but you know it's it's something hopefully it's not going to be another e-mail that people will start ignoring because exactly every single day. Exactly. That's always the concerns of these things. I remember my whole time when we we we just how many alerts this night, tonight ignore everything's running. Yeah. Oh gosh. So next one, I need some sound effect. Oh yeah, yeah. And this one's really cool as well. Not the sound effect, the bit of news 

commitment. So that was a discovery done by John Bryant in where I work. So there is a new RDS instance stuff, which is the R7G. So it's a graviton one. And what we notice is that the RDS one does not have. R i's so you cannot buy an RI for an R7G 

DB and I've I've looked on the calculator with Postgres, definitely not there. I'm pretty sure it's the same for any other. So is that a sign from a WS that they don't have the capacity at the moment for R7G or in all the real or something like this? I don't know, but that was quite interesting to discover that sometimes there are no R i's. Very surprised. I don't, because normally our eyes are released with an instance. As soon as something's there, it's only there. But the pricing doesn't exist in any website. It's not on the pricing website, it's not on the API, so it is very interesting. So maybe it's just, well, you can't imagine it's an oversight because it's such a standard piece of their world and they release instances all the time. Yeah, but I I don't know and even the the inside you have this. So I was looking for the pricing, it shows yeah utilization on demand only it's when you go. So when you go to the pricing model, you will just get the you would just see on demand and I was that was shocking which is why we highlighted and we are liking here to you guys it's just to say it might be just a mistake, it might be something but. Hey, like we discovered the 5% increase of a generation of a generation. Now we found you the first instance type that doesn't have our eyes at the beginning. Yeah, 

right. And then to my next, this is more just like a nice little thing that we found out and it was you can now download the savings plans. Price sheet for Azure and you can find this in Learn Azure Cost Management Billing Azure Savings Plans. You can find it under there and you can actually download the EA price sheet and the MCA price sheet. Which is really useful because actually it can be a bit harder to find out the pricing I think or just because of it being quite a new change and all this stuff and it doesn't automatically take over the pricing of something else like it was convertible RIS for AWS. So now. That's there, it exists, and it walks you through how to find it so you can get the full price plan records for whichever service you are, or whichever flavor of Azure you're on. And is it me or I remember that Azure every customer has different price plans. So this is going to download savings plan. So, so if you're on MCA which means you're either pay as you go or on CSP, then no, you're on. They have a single driven price book led from from dollars, the EAS, they were holding 28 price books. Based on currencies, but now they are changing that and that's why they had the 9% price change because that was an FX adjustment. So what seems to happen with that with EAS and this is NDA but I'm not dealing with any. So this is just knowledge from that. I've just found out without ever being put on NDA for that you know you get a then a discounted like a bit like an ADBS EP or a Google commit off that price book. So I think it's still centralized price books then with some negotiated pricing discount changes to those. So I would imagine on your EA price sheet that calculation's already done for you. You would think, I don't know for sure, Okay well as usual. Listeners, if you know something I want to share it with us. We are working hard to create new communication channels like on LinkedIn, YouTube and all this kind of stuff, but feel free just to try sending us an e-mail or comment on 

wherever you find us. Now again, music, please Maestro 

Savings and I am on Azure so this is GA spot priority mix. This is so Azure is excited to announce. I am happy to announce that there is a general availability of Spot priority mix which means that you can mix now into the same group of scalability both SPOT and standard Vms and then it will just start trying to use. I guess my guess would be starts going to try to see SPOT and if SPOT is not available or if SPOT are being recalled it's going to start normal Vms but you can now have. Both together, you can choose a base number of standard Vms, a percentage split of spot and standard Vms to be used then for scaling and so. That's quite cool because finally you don't have to choose one or the other, you can just mix and match. 

And I remember us talking about a similar release which I think must have been maybe December, November because it's not in this document because we do it by year with all the news on four AWS where we talked about and we thought it's pretty cool. So it's obviously been seen as something that is worth having across the piece. Next one I have, I don't want to go into too much detail about it, but is. That you can minimize. Code starts with recommender on Google and so it's a bit tricky because this is release note stuff, but the Cloud Functions minimum instance recommendations are now available in preview on recommender. And you can also reduce your cold starts with this release where the text function that could benefit from setting up minimum instance sizes. In other words, functions that are likely to be negatively affected by cold start issues informs the user about the opportunity to eliminate the cold start issues and lets users set up the minimum instance directly from the recommendation system, which is quite nice. You don't have to then go out to it, you can click and get it sorted from there and it's automatically available the recommendation after. Deployment for a function as soon as it's been running for a week, which I think's really cool. And so yeah, that's a nice new touch. It was a bit tricky to find, but at least we've brought some news from the Google world. I think that's a cool thing to be talking about. Yes it is. I don't know why they don't do blogs about them, make my life easier. 

It's just because they want us to do search and dig and find them. They want someone to read their release notes, the amount of work they put into them. We've got one person reading that stuff, but yeah, that's 

it. Yeah, so we've got. Now we are into the storage thing, Storage section. I have two S 3 News. One is GA, so it's AWS Data Exchange for Amazon S3. With data exchange is a way to either find subscribed third party data file. So for example, and I think that started when there was COVID that they are, for example, some hospitals or some big statistics that everyone wanted to access. And you could say, hey, I have interesting data there. You can put it into an S3 bucket and you can start. It's also in a way either monetize it or at least making it available to other people in a structured and controlled way. So that is quite cool. It's, it's available, it's a GA. So have a look at that In the other it started with things like IMDb, which I think a lot of people didn't realize Amazon owned and they basically made all that data available and it was basically it started being a way of them being able to sell their own data and then it stretched out and yeah, it found some really good uses in code types. Nice. Didn't know that the second news is introducing Mount point for Amazon S3 high performance open source file client. So the idea is you they create, so this is an alpha release, so very, very early you might have anything could change, everything can change. You might destroy whatever data you're accessing. So yeah, don't use it in prod yet. It normally will allow you to mount. So if you're coming from a Linux file system world or Unix file system world, it means that you can have it just as a folder. You can go into a folder onto Unix system and it's going to be. Listing stuff that is in S3. So all of a sudden it's possible to copy and move things without having to rely on the AWS client For example, it is managed directly by the operating system. That is quite cool. I want to try it out, but it's an alpha, so maybe we'll see. But 

there is a GitHub, there is a technical blog, there is a roadmap, You have documentation yet it is an alpha. So reading easy, Probably writing 

not, because I've got one of those beautiful ones that occasionally Microsoft do where the entire thing is 6 lines. So. The reading up was not a lot of work and that is that you can now do more transactions at no additional cost for Azure Standard SSD. So they talk about as part of their commitment to deliver the best possible value for Azure Disk Storage customers, we're announcing a waiver for billable transactions that go beyond the hourly limit for Azure Standard SSD customers. As a result, we made changes to the billable transaction costs per hour that can result in additional cost savings. But did you hear the? The magic word in there can so where this was multiple costs split out. They've now been merged, and I imagine maybe the merged one has gone up a little bit so you could save. 

So I haven't looked into the actual costings of it, because then you have to go and find old data to compare with the costs going up and stuff. And while we do a lot of stuff with Amazon costs and billing data, because that's easy to get and handle, some of the Azure APIs are a little bit harder and still playing with. So I don't have background data. It might be worth asking our friend John Bryant if he's got any. 

Ah, he always stuffed there. He's always he's got incredible Machine right. Well I'll do the thing. But now we've gone to some cool things that you've been seeing. 

So that's yeah, misk silly. It's so recently I've been playing again. I went back to data around my core because I was getting slightly bored of doing non tech stuff. Sometime I need to go back to tech. And so I did some work and one of the things I updated Windows license cost on AWS, how much do they cost? And Red Hat also license cost on AWS I used a tool which is called Observable. So if you go to observablehq.com you can search for. Probably if you search Windows license you'll find me or RHEL license you'll find. And but the idea which was very, very interesting for example, is when you put the numbers. So the numbers that come out are quite interesting, but you end up like saying a Windows license when it's provided by a WS cost you something like $400.00 per VCPU per year. That's a lot. So initially I thought it we would have a massive capacity from a WS to negotiate really good good things or whatever, but it ends up by being. In my opinion quite expensive. So I understand now that there is an option which is BYOL bring your license to all of this. Red Hat is the same, I think it's 500 if you have less than 4 CPUs and 1000 if you have more than four CPUs per instance in this case. And yeah, so I've created that into. You can go to the article. You can also go to YouTube channel. At the moment it's fully branded. Me, but I hope to change that and make it fin up guys branded. So that's going to become and every research or every news that we found interesting or presentation we make it public. We'll try to record that and put it there. So not only we have now a podcast but we'll also have some videos for you indeed also just for those listeners in the UKI think there was a great, I couldn't make it but there was the did you go to the Shard meter. Yes, I did. I spoke there and I recorded the thing which was fantastic. I did a presentation there 

about who cares about finops, Nice. I like that. And then there's one in the leads coming up as well for those in the North End, which I will try to get to childcare dependent. So yeah, which I get more and hopefully we'll be able to see everyone that's going across the Finops X as well, yes. And by the way, I think there is a finops event quite that. Our good friend Mike Bradbury is organizing Manchester. I would be. There's another one in Manchester as well I think yes. Yeah we've done we've done one. Let's see if I could come up with that one to see you. You can state mine yeah, you can state mine. Not trying to say because it's we're not that far Steve and I okay. We're just a train. It's some hours. And you see, the only place we're going to meet this year potentially is San Diego, which sounds a little tragedy. Absolutely mad. Yeah. I feel like even when I come down to London recently, it's been for a meeting. So I've been down there for like 2 hours, like 4 hours to do, do, 2 hours about to start saying a bit less with these things going on at home. But there we go. I don't think there's any more news and this just tells you how. How much of A chat this is between me and frankly, we're actually just talking about our our our socialize and when we might be able to get to events to see each other. But hopefully we'll get to see some of you at these events that are coming up. I'm really excited for FINOP sex and some really cool stuff. And yeah, there is also Roadshow, I think, yes, also the Phoenix Roadshow, if you are there, just look for. So for me, look for the red glasses. If you see a guy, bold guy with big red, round red glasses, that's me. Chad, very happy to hear from you. It's always good to know new people and to discover that. Yeah, we have listeners. That's super cool. Love to talk to you guys. Yeah that would be true. I will see if I can get down. I was going to try to get to the Amsterdam one but I can just life stuff you know it's like at the moment company family fun 

times the podcast getting harder and harder but it's it's an important part of my my every two weeks and hopefully everyone's had a chance to listen to the Dean of Solace podcast which I found really really fantastic to do. We've got some cool new ones coming and thank you people that have been contacting us to to. Come on and talk. If you've got a topic we'd love to have you on. And yeah, that's super right. Well, I think we should probably wrap it up, Frank, because we need to go and do our real jobs. But everyone, thank you for listening. Thanks so much for your help as always. Speak to you soon.