The Cloud Pod: A Masterclass in Cloud Architecture - Episode 56

January 31, 2020 00:47:15
The Cloud Pod: A Masterclass in Cloud Architecture - Episode 56
The Cloud Pod
The Cloud Pod: A Masterclass in Cloud Architecture - Episode 56

Jan 31 2020 | 00:47:15

/

Show Notes

Your co-hosts move from the atmosphere to DigitalOcean as they recap the week in Cloud on this episode of The Cloud Pod. A big thanks to this week’s sponsors:

This week’s highlights

View Full Transcript

Episode Transcript

Speaker 1 00:06 Welcome to the club, Justin and Peter, Speaker 1 00:19 episode 56 recorded on January 22nd, 2020 the cloud pod, a masterclass and cloud architecture. Good evening. Hey Justin. Hey, how's it? Uh, how's it going? It's it a lovely Wednesday for both of you. How am I going to say? It's a lovely Wednesday. I feel like I've aged about 10 years this week so far. And it's unfortunate. How about you Peter? How's your week going? A spectacular. Uh, I have not yet finished, uh, even considered starting packing for my trip tomorrow, so, Oh my, uh, my wife is also leaving on a trip tomorrow. Uh, not with your Peter, but uh, so, uh, but she's frantically packing as I record the podcast. She is behind schedule and she is feeling the stress as she's leaving tomorrow at like 1130. We'll see how bad we'll see how that goes. A long night as my station. Well, good. Uh, we got, we had some followup. Speaker 1 01:11 Uh, so from episode 54 and the Justin does a thing segment. Um, I think I said something to send of, uh, reading through the Oracle Prosser announcement that I didn't really see where they mentioned, uh, that they had EBS volume black coffee before. That's kinda what led me down the path to the Oracle thing. And, uh, so max Varun, product manager, Oracle, uh, he sent us an email saying, Hey, uh, we did actually mention it in the article that you link to, uh, in the second paragraph. And, uh, it says in the second paragraph, cause I went back and verified, uh, this new capability follows our prior announcement of cross-region block falling and backups together. They provide a complete solution for comprehensive application and data protection in the cloud and cleaning the easy recovery of compute instances across regions. So yes, technically it was there. Uh, but if you ever have to go through several hundred press releases a week, um, you've skimmed through the, the open boilerplate because it's just boilerplate and I just missed it. Speaker 1 02:03 So, but it was still a fun adventure. And so I, I don't regret it at all. That's right. Often that, that now, so the original aspect came right before we started recording the podcast. It ended up in between the 18. So, uh, I think we can forgive you for missing that one. Yeah, that's all right. But it's still fun time and, uh, he actually wants to talk to us and you know, get some more of our feedback and as well as maybe do an interview show with us. So we'll, we'll see if we can get max on the show and talk about it. So I got to say, I got to see the email too. And he was very polite. He was very polite. Yes, he was super nice. And you know, and I, I was listening back to it, uh, earlier in the week or late last week and I thought we were pretty nice. Speaker 1 02:38 I mean, I wasn't mean. I mean, I believe the gentlemanly size and you know, this is a, you know, very infrastructurally focused, um, solution, but it works. And the ice guys, he's great. And you know, you know, if you want the things that might've come, I think the only negative thing I said is I didn't think it was very developer focused. Uh, it was more system engineer, more infrastructure focus. But I think that's, I don't know if that's really a complaint as much as it's a, it's just a fact. So, uh, I think it's just the way it is. And there's people you build clouds for and those are not those people. Speaker 2 03:07 Cool. And don't forget Oracle support to get back to you, uh, almost two weeks after you had the, had the payment this year Speaker 1 03:13 we recorded that episode, uh, on, uh, January 8th and I think I tried to get onto the Oracle cloud on the sixth. And so remember I mentioned to you that the credit card process failed and it was some issue with their backend system, but they had a chat with an agent and then it emailed, had option to email. And so on Monday, on the 20th, uh, exactly two weeks later, uh, they basically said, you should clear your cookies and try again. Uh, in an email response back to me, which I was, uh, not super appreciative of cause it's the eighth in two weeks. And so I, you know, if I had not got a Twitter and I had not complained and use my Twitter credibility to get that fixed quickly, we would still be waiting for a Justin doesn't thing until this week. So. Speaker 2 03:57 Well, Hey, at least they got back to you. I mean, if it was AWS, if you don't spend, you know, 10 K a month, they won't even speak to you. And that's true. That's true. Speaker 1 04:04 Let's go and talk about Microsoft. So Microsoft is planning to go carbon negative and this raises the bars for their tech rivals. Uh, this is a 30 year plan for Microsoft to remove the app from the atmosphere. All the carbon has ever admitted from the founding of the company, in fact, a effectively eliminating its global carbon footprint. Um, other companies have set goals, but nowhere near as ambitious as at Microsoft's. And by 2025, they will shift to renewable energy, to a power to facilities around the world. And after that in 2040, they'll start going towards carbon negative by planting trees and forests and doing all kinds of things to actually give back, uh, carbon to the world that they stole. So there you go. It's engineering investment, uh, alphabet. You know, as previous report, did they reach carbon neutral in 2007. Uh, they have not committed to going to carbon negative and Amazon, uh, you know, committed to some goals that we've talked about in the show in the past, uh, with desire to be carbon net zero by 2040 and a hundred percent renewable by 2030, but there's been reports as we've talked on the show before, that they are way behind their goals, uh, deliver that as well. Speaker 1 05:04 Uh, so Microsoft is making a big bet, see if they can deliver on this promise, which may be a bit ambitious. So they're going to continue to grow the Asia cloud or the at the level they think they're going to. Speaker 2 05:12 Yeah, I'm not sure this is a fair comparison, really saying, Hey, look, look what we're doing. And you're not doing the yet when they've got 140 550,000 employees and obviously a much smaller data center presence than Amazon has right now. And not to mention that Amazon is not just AWS, it's the entire Amazon business including, um, you know, all the, all the trucks that they deliver with and all the warehouses that they support and they've got 750,000 employees. I think it's not quite affect comparison. Nobody fed life is fair. Speaker 1 05:45 It's all about, it's all about those marketing eyeballs. And if you can throw your competitors under the bus and say, we're going to do something more amazing and then hope that everyone forgets about it by 2025 and 20, 30 and 20, 45 and all these years, many, many years in future when we all will be drowning and global warming, you know, we'll deal with it then. So Speaker 2 06:02 it's really cool. Um, I joke about that stuff but it's really good to see that such as a profit focus company is, is spending so much effort on doing the right thing. Speaker 1 06:13 Definitely good to see these things. You know with big investment companies also going down this path of you know saying they're going to start investing more in carbon negative and carbon neutral solutions. It's definitely a changing world in this area. You can wave goodbye to some people at digital ocean. They've apparently laid off about 10% of their workforce. It was undisclosed. What is estimated 10% and they, it's interesting cause on hacker news co-founders actually had a few things to say about the recent layoffs. So first of all from the register article, apparently the digital ocean was making 200 has a 275 million in annual recurring revenue and about 500,000 customers globally. Now I don't know how to define a customer because I might be a customer with my one server versus another customer that has thousands of servers. I don't know how they treat those, but they do say the new org structure with the, with a 10% reduction in position to accelerate their profitable growth by continuing to serve the developers and entrepreneurs around the world. Speaker 1 07:07 And like I mentioned on hacker news, digital co-founder Moisey Jarecki pry a little bit more detail and then admitted that he did not have authorization to say those things, but he felt they were still fair and left them on hacker news. So he says, uh, as unfortunate as layoffs are, they were really due to the two CEO changes in the past 18 months and leadership changes that created some competing directions in the business, which antsy Spruill our new CEO is now addressing, we're not running out of money, nor do we have an immediate need to raise capital. And the layoffs aren't laid to any of any sort of cost cutting initiative. So interesting to see what kind of happens and if this is truly not a cost cutting move and truly designed to streamline their business and focus on profitability or is this a first step in getting acquired by a bigger player? I think that is still to be determined, but, uh, definitely good to see, you know, Moisey Jarecki out there kind of defending the company he founded. I think that's a great, Speaker 2 07:58 it's a shame they couldn't keep the PayPal on then just to get them all working on something that was more profitable or uh, heading in the right direction. Then let, letting good people go by definition layoffs and her cost cutting regardless of whether or not you're running out of money or raising capital. If you're laying them off, not with the goal of reallocating that money somewhere else, then, uh, it's cost cutting. But yeah, I mean it makes sense if they're spending too much today. Yeah. I assume that they're big enough to be, um, subject to restrictions on, on rehiring after layoff too. So it's growth for a couple of years. That's, that's concerning. Maybe, uh, maybe your vision of aggregation of this, the smaller providers will actually happen this year. Speaker 1 08:41 Just another speculation here in the, um, you know, in the article from the register that I didn't have definitely show notes, but they do mention cloud providers like digital ocean, Linode and vulture have been under pressure for the industry zone and player Amazon since they launched, uh, the low cost offering called LightSail back in 2016, which I never met anybody actually uses LightSail so I'm glad there's, apparently there's competitive pressure from that. Uh, but, uh, it was a little interesting as well. And then you also did mention that they are running at a modest loss right now, which is okay because they're growing and OS because whenever you launch a new product or feature, the upfront costs are as much higher to get the initial pub product built and there is no revenue contribution from until it's launched and ramped up. Uh, that again was from their founder. Speaker 1 09:20 Um, so they haven't raised money since 2017 and he says that they have no plans to do that at this point, uh, to raise on around anytime soon. So we'll see what happens. But, uh, I do still expect consolidation in that space to finally fulfill my 20, 19 prediction. Well, someday, someday it'll happen. All right. Moving on to our good friends, uh, up in Seattle and Amazon web services. Uh, they, this last week was sales kickoff for them, so most of them are out, but they did not slow down the announcements. They kept us going with lots of great new features. The first one is the ability for the eight of us health capability to enable aggregation of health events across all of your AWS organization connected accounts. The central aggregation of these AWS health events allows you to see a realtime access to all of the health events, including visual accounts and organization, operational issues, scheduled maintenance and account notifications and these or view unlocks new capabilities not previously available to you. Speaker 1 10:13 Use a customer, including central ops team's ability to view and respond to events affecting accounts across the entire organization. So a great, continue to see a lots of Amazon organization's features being developed. Um, after this got announced, I did a Twitter, a tweet at AWS, and I said, uh, you know, I really liked that you have these public roadmaps, but will you please put Amazon organizations on a public roadmap so I can stop coming up with solutions that I want to build that you already building really wouldn't be any good to have this right before the RDS certificate mess. Yeah. So that's been really good. Yeah, this is good then because thinking about a hardware degradation and machines AP rebooted and stuff like that, it's very difficult to keep track of that over a couple of hundred accounts or mall. And so putting us into essential places is really cool. Speaker 1 11:01 That's the way to do this before was to kind of build our own custom tooling and you know, puck call the API's across all your accounts, the trusted advisor, and then integrate with your ticketing system. So this is all very, very nice. Yeah. Uh, Amazon, uh, ECS has announced the previous support for the EFS file system. Uh, this will allow you to basically have stateful data be stored on EFS and connected to by Amazon ECS. Uh, this is configured in the ECS task definitions and it's compatible with the easy to launch type. Uh, customers can add the file system to the task definition and mounted as part of the launch process. Uh, EFS previous support, um, can be set up through the task definition with the volume ID, uh, using the EFS file system it and assume that this is probably the first step towards far gate being able to support EFS sometime later in the year, which has been a much, uh, requested feature. So good to see this step in ECS, which means I might not be far behind for far gate. The announcement on the I M controls that EFS makes all the more sense. Now it does make a lot more sense. Speaker 2 12:00 I still think the S S three would be a better solution than AFS. Certainly a lot cheaper and probably as performance. But I'm, I guess this, this is a bit of an enabler for people moving legacy applications that still need, uh, to look at what looks like a file system to them. Speaker 1 12:15 As someone was mentioning, I think it was Benkei ho actually on Twitter was saying, you know, he'd love to see EFS to kind of become a caching layer to ask three <inaudible> kind of interesting idea. Um, that can be really powerful for a lot of companies who are wanting to make the pivot to the lower cost S3 capabilities, but without the overhead of rewriting into object storage directly. Maybe it already is that, I think it probably is at some level, but you don't, yeah, if you don't get that pricing opportunity, cause EFS is super duper expensive. Yeah. I don't think it is. I think it's UBS packed. We'll have to do some digging. Yeah, definitely find out why it, yeah, sure we can. I'm sure we can find a product manager somewhere who might tell you. Yeah. Yeah. It's just a matter of hitting the right people up on Twitter. Speaker 2 12:57 Hey everyone. Jonathan here. I just wanted to take a minute to thank the cloud consulting gurus at Fargo and for helping make the cloud pub possible. These folks truly get it. Cloud consulting experts since 2008, they are premier tier partners with AWS, Google cloud platform, silver and Microsoft is your partners from multicloud to containers to moving for production workloads to the cloud under the tightest compliance vocals team of full stack cloud engineers have been there, done that, gutting the tee shirt and are ready to share their experience with you if you're in the market for some talent. To supplement your team, visit Dolly, Dolly w. dot. Fog up STO IO slash the cloud pod, www.fugups.io/the cloud pod for corn. The promise of cloud delivered Speaker 1 13:43 Amazon EKS has announced a price reduction. This is a 50% reduction of the per hour cost for the EKS cluster. This is for the course of the managed master nodes. Uh, this new prices available for all new and existing Amazon EKS clusters. Uh, this price reduction now brings them to exactly where they were before, which is still 100% more expensive than EKS and GKE for the same capability. Uh, but they did want to point out that, uh, the ECAT feature has really 62 new features and launch in 14 regions and for and support for four different versions of Kubernetes, uh, in the last 18 months. Uh, which apparently supposed to make you feel good about paying this monthly fee. And they did highlight three customers using mission-critical production workloads on EKS, including a snap, HSBC and Volo and, uh, Edward <inaudible>, uh, lead SRA at Apollo says we are running our application on the Amazon EKS launching up to 2000 notes per day and running up to 75,000 pots where microservices and machine learning apps allowing us to detect purchase intent through individualized, individualized marketing and the website and shops of our customers. Speaker 1 14:41 Uh, that is a lot of pods. I had a lot of pods. I would not want to do a, I imagine it's SRA. He is a very busy man. Derek in this room did point out though that GKE doesn't auto scale a right size, the uh, the cluster, whereas, um, EKS does, it will scale down the cluster or scale it up as necessary for the workload. There is a GKE premium offering that I think does do that, but again, it's additional pricing for this capability. So I mean, if you really get these comparing comparisons down to, you know, GK premium versus Amazon EKS, I think they are probably in close to parity at this point with the new price reduction. Um, I still would love to see this as free, but I do think they kind of see the, that's what ECS is and far gate, uh, you get all those benefits of containers without the overhead of Kubernetes. Speaker 1 15:25 And so if you want it for free, go use the native tool. If you want Kubernetes to be a cloud agnostic, you're gonna pay a little bit of a tax. That's that. I've heard zero customers complain to me about EKS pricing, having managed Kubernetes masters, I would pay it every day of the week. I was happy paying it volt price. Uh, but I mean, if I was multi-cloud and I was looking at Google and I was looking at Asia and I'm in a competitive situation against Amazon, um, that is a cost that gets factored into the cost, the ROI story for these systems. Absolutely. Amazon also decided to announce a additional price cut this week. So CloudEndure, uh, which we talked about on episode five going way, way, way back, uh, when it was acquired by AWS. Uh, and again in episode 28 when they discussed that the cloud and migration feature was going to be free and available to professional services, customers and migration partners to help you move to the cloud. Speaker 1 16:14 Apparently this feature, uh, is a DVR automated der capability that allows you to replicate data from on-premise, virtual or cloud based systems to a low cost staging area and AWS region of your choice. And then you can use that data to then extend, create automatically your production environment. Uh, so apparently that now has an 80% price reduction. Uh, they'll say is cost reduction now reduces this to about 0.02, 8 cents per hour or about $20 a month per server. Uh, for the stage and capability. I assume you're also still paying for EVs. Uh, our S three storage. Uh, but again, compared to a traditional der, the amount of this cost is pennies. And so this is a nice option for a lot of companies who have to have a dr solution but don't necessarily need to have an active active type standardized solution. Yup. Well on its way to free on its way. Speaker 1 16:57 Yeah. It's odd that they didn't build this replication end to the EBS layer rather than having it like an agent based thing. But maybe, maybe it just works better when you've got an agent in the LS doing the work for you. I imagine it's, it's better. Uh, and plus this is a third party they bought. That's some level you'd add to make a decision. Do you want to re engineer this product to really truly tie into the Amazon ecosystem? Or is this a, a short term solution for a specific use case that you know, people are willing to pay a premium price for a, and you can deal some of the trade offs. I think that's what they made the call for. Well, I guess now they own it. They have zero incentive to do it at the IBS laid because of what else solution it works. Speaker 1 17:32 And why, why reinvent the wheel? That's true. We talked about at reinvent that uh, Amazon has released their new local local zone in Los Angeles and we commented then that it was a little weird because they have a local region. And what's the difference between your local region and a local, uh, but uh, Amazon is a decided to announce to, to address that issue by announcing that a Amazon, a SoCo local regional now be expanded to become a full region. Uh, so of course when you can't solve this problem with marketing, you just buy your way out of the problem by building two more availability zones. Uh, and so they will now have three availability zones in early 20, 21. Uh, and this will be the second full region in Japan after the Tokyo region, the Asako local region was launched in February 28 and uh, that is about 400 kilometers or 250 miles apart, uh, for these two data centers. Speaker 1 18:18 So, uh, by 2021, you will have two additional availability zones springing your toll up to three for the local region in Osaka. There you go. There is no more local region. It's only local zone. Yeah. Now you can do full der inside of Japan, which is nice. Um, unless a title wave takes up the whole Island to then you don't really have a lot of options. You have a lot of problems. Then at that point too, that's all it takes you out. I don't know. There's other media that you can do at that point. Yeah. I thought as long as they're 50 miles apart, it doesn't matter. They're turning 50 miles. God. So looking kind of straight across Japan and minutes of the West. Hope for the best guess. So. And then the Amazon key management services is expanding support for asymmetric keys. Uh, this allows you to create asymmetric customer master keys or cm case and generate data, key pairs and all regions where KMS is available of course except China. Speaker 1 19:08 And this feature enables Amazon customers on third parties perform unauthenticated encryption outside of Amazon on premise or in an application or something, and then provide that data to you and the cloud. And you can then use your private keys with authentication to decrypt that data in the cloud. Uh, similarly, customers can use ECC or RSA private keys generate digital signatures and third parties can perform a verification outside of Amazon KMS using those public keys. So this is a great way to kind of get some portability of signs of signing as well as, uh, encryption outside of the Amazon world that you can then decrypt inside AWS. Sorry. Great. We've got PGP for rough. I came, miss like has happened is released. That's awesome. No, seriously, this is, this is good cause this, this is like a huge constraint on KMS, which was saying the number of concurrent operations, if you no longer need to keep half of the key, you know, the public key and KMS or any cushion key and KMS and you can, you can just literally put it any website, I haven't flushed it out in public, then you could scale encryption infinitely. Speaker 1 20:09 Whereas as before U S somewhat limited by, uh, by the KMS pod limits. Yeah. And also it's free. It is. I mean other than your, your key, you're storing KMS you to pay for it. Other than that, yeah, almost free. Oh man. So again, Asia is still living a little bit quiet. We did just check before the recording and they have not released a new feature since, uh, December 19th. Uh, so again, they're, they're still a little busy, but in conjunction with the announcement that Microsoft is going, uh, cloud basically carbon negative, they have also released the Microsoft sustainability calculator to help enterprises analyze the carbon emissions of their it infrastructure. Uh, and when you really dig into this, it's a power BI application that you can install that allows you to pull in your Asia data to let you know what the sustainability, um, situation is of your Asia cloud as well as you can use it to. Speaker 1 21:00 You can send a little bit to start pulling in data from your on premise data centers, um, and determine what your cl, your eco footprint is in the world. Um, this of course is all in an effort to get you to move more of your though to Asia so you don't have to worry about making yourself carbon neutral. And this report will help you, uh, make, make this determination about what you want to do for carbon based offsets. I would if Azure is going to announce some lower, uh, low power processes cause we talked about running windows and the sequel service. So probably not. Yes. Uh, as I was just talking about moving a instance from in five 24 X large to our five 24 X large because of the memory hungriness of SQL server. Uh, yeah, I don't, I don't know that low power is gonna help them probably, well, hopefully the, uh, Asia marketing people, um, start announcing something soon. Speaker 1 21:50 Uh, I have, I mean, in fairness to them, they release a ton of case studies and things that we want to fully cover here on the show unless they're super excitingly interesting. Uh, which none of them were. Uh, but you know, hopefully they'll start releasing some stuff soon here. I know they have some ignite conferences coming up, some local ignite conferences. Uh, so hopefully it will start saying a couple of things maybe come out of those here in the next few weeks. Uh, cause, uh, I've seen that they've been awfully quiet or, uh, maybe they're up to something else that's more important, that totally busy working on their ultra premium advanced storage systems. I mean, I do think you commented earlier that they're, uh, they're working there. They're trying to, uh, uncramp themselves from <inaudible>. Maybe that's what they're doing if they're gonna, if they're a little busy trying to get the genetic contract figured out. Speaker 1 22:31 But, uh, there we go. All right. Google has announced the launch of their premium support offering for your enterprise and mission critical needs. Uh, this launch for the premium support will include a robust set of services and systems to serve enterprise and mr. cruel needs of the Google cloud customers. Uh, if you elect you get the premium support, uh, you will have several benefits including your cases will be handled directly by a context of where expert who understands your unique application stack architecture and implementation details. Uh, the team will work hand in hand with your Tam to deliver customer centric support experience with faster case resolution, more personalized service and higher customer satisfaction. And premium support, uh, also brings system between GCP andG suite, including a more competitive set of new features and services including simplified pricing, uh, in addition, if you'd like to pay additional money to your premium support. Speaker 1 23:14 So premium support plus plus, uh, there are add on packages that you can add to this. Uh, and those includes the ability of the advanced event management service for deeper architecture review and increased readiness for peak events. Uh, expanded Tam coverage should address multiple time zones. And the mission critical support, which is in pilot with customers and available later this year. This service will help you, uh, help customers evaluate their dev ops SRE practices and suggest recommendations though improve supportability and reduce downtime for mission critical workloads. Uh, so that is the offering. What do you guys think the can who are experts who understand my unique application stack? Sounds uh, Speaker 2 23:46 sounds wonderful. Uh, Speaker 1 23:48 great. I don't know how you actually do that, but I appreciate the effort. Speaker 2 23:52 I mean, I understand your stack and knowing your stack in advance when you need them to. I suppose it's, it's two different things. So I guess they're going to have a bunch of smart people Manning the, uh, the service desk. I mean they just overstaff in the beginning when they have very few customers signing up for it and then, um, figure out how to scale it later, I'm sure as their plan. Although it's surprised that they're coming out of the gate at a higher price than Amazon, which is a, it seems like when you're not in first place and you're have the me too feature, it's time to, uh, immediately at least match pricing. Speaker 1 24:27 Well, and even even some of the things that they're, again, the add on like the advanced event management service, Amazon provides that as part of their premium support today too. Um, that's already included. It's part of your fee. Uh, the other two, the expanded Tam coverage, that's actually a really cool cause that is what challenge, if you're a global international company, your Tam doesn't work 24, seven. And so if you're having an offshore team in India that needs to talk to the Pam, um, they're not available. Um, so it's nice to have at least have the option. I wish Amazon would give you that option as well. And then the, uh, the mission critical support is really just a way for them to kind of take the, um, SRE dev ops practices they have. And really they've had a consulting service that you kind of get all the time. Speaker 1 25:04 And available to you. So it's kind of a nice add on. I see why those are additional prices. But yeah, when you think about you're actually getting a little bit less potentially and capabilities, but you're also getting a little bit more specialty and your infrastructure. Um, it is a little weird that it costs 150 K your base plus a 4% of your net spend. Um, now the, the first part of it is in line with Amazon. So if you're a few, if you're spending less than a certain number amount of money in a, in a month, uh, you'll basically pay $15,000 a month. That's a minimum of $150, 150,000 a year. So it was equal. Uh, but that's for the up to 150,000 a year. Then you have, uh, but Google's already charging you 4% more on top of that for even the first tier. And then from there it goes up significantly. Speaker 1 25:47 So I modeled this out in a couple different ways. Um, so I was just like, okay, what if I'm spending $100,000 a month? Um, that's basically, you know, $16,500 per month on GCP. It's 15 K on AWS, cause you're not crossing that minimum threshold. You jumped to $1 million. And I thought, you know, as you guys got larger that, you know, the discount tiers come into play and maybe it would make more sense in a larger account and they're really trying to price us more towards larger accounts. But even $1 million a month in spend, you're at $52,500 on Google, uh, versus $30,000 on AWS. Uh, which is really kind of shocking to me that that price didn't scale the way I expected it to. Speaker 2 26:20 Huh. That's interesting. So if you get the enterprise discount, um, do you, do you pay support on the list price or do you pay support on the discounted price Speaker 1 26:28 for AWS? You pay the discounted price, so they give you a, you know, whatever your discount is on your EDP. Speaker 2 26:37 Yeah, I guess, I guess when you're spending two or $3 million a month, it's uh, the difference is negligible to them. Speaker 1 26:43 Yeah. It really kills my argument though cause I've been, I've been kind of having this fight with Amazon that um, I'm not getting the value out of my Tam that I really want at the price point that we're paying for them. Uh, you know, cause there's, once you cross that 15,000 a month, which was the basement on pay, which pays for a Tam, a Tam is shared across, you know, six to eight accounts, I think. I don't remember exactly number of counts, but you know, I'm saying, you know, once you go over, you know, you're spending almost 800, $900 million a month. I'm spending spending way more than 150,000. I should have two Tams. And I'm trying to make that argument and now, now I'm supposed to be like, well, you know, look at GCP, it's more expensive for less. Speaker 1 27:20 So you're getting a steal of a deal, Justin. That's how, that's my, that's my fear. But hopefully that I'm listening to the podcast or paying attention to a JCP costs civil customers, they've got to listen to podcasts. And indeed, Google has been doing a pretty good job picking up customers and we'll talk about that in a little bit. But they also apparently have not won over the hearts and minds of Epic systems. And you probably are familiar with the name of it. If you've ever looked at the computer at your doctor's office is typically an Epic systems, uh, patient record system. Uh, apparently, uh, according to this article from see NBC, uh, account representatives from Epic systems have been calling their customers were using Google and letting them know that they will not be pursuing further integrations with the Google cloud. Uh, Epic will instead focus on AWS and Azure. Uh, they decided to halt development with Google cloud because it wasn't seeing sufficient interest among its health system customers to warrant the investment. Of course, this comes as a big blow as a Google's efforts to find new customer segments where it's cloud product. A, as a company lags on AWS and Asia, the a company I'd recently signed a bigger agreement with the Mayo clinic, uh, which I thought would have maybe helped us out, but apparently not enough to prevent Epic from saying there's just not enough business to make that spend work. Speaker 2 28:30 And I find it a bit weird that they really care what cloud their customers are working in. I mean, you think that any integrations that they built into their own product would be, would have their own API APIs and be sort of cloud agnostic. So it's, I F I find it a little strange. I mean I, it makes sense with as your going go and kind of all in on the, um, on the patient records systems, uh, last year and Google not doing that quite so much, but it's bizarre that they're telling their customers basically to not not use GCP if they want support from Epic. Speaker 1 29:01 Well, they're basically saying, you know, basically Epic is saying we're not going to integrate with GCP. So you, you can still go use GCP all you want to for medical imaging and AI and machine learning. All that, just your add or build your own bridge between the Epic system and Google. We're not going to do it for you is how I read this. Um, I don't, I mean I don't, I never deployed it or supportive Epic, so maybe, I don't know. Epic's completely SAS these days or if it's a on-premise, um, type setup where you know, you run it in your hospital infrastructure either in the cloud or on prem. Um, I don't really know how Epic works from that perspective, so maybe it's a combination of all the things. I don't really know. Other couple interesting quotes here. Epic's a vice president of research and development. Speaker 1 29:39 Seth Hanes said in a statement we substantial time and engineering effort in evaluating and understanding the infrastructure Epic runs on scalability, reliability and security are important factors we consider when evaluating these underlying technologies. Uh, he said Epic focuses on supporting infrastructure the Epic community uses today and is likely to use in the future. Um, it was also noted in the article that the wall street journal had reported that Cerner, uh, had been pursued by Google, but ultimately turned down over tens of millions of dollars in incentives and went with Amazon instead. Uh, and there's an article here from Aneesh Chopra, president of health technology company care journey, and he says, uh, we've historically seen hospital systems make these decisions independently of their medical record provider. Uh, it'll be interesting to see if Epic's thumb on the scale moves cloud market share in the healthcare space. Yeah. So it'll be interesting to see. Speaker 1 30:23 So it does sound like maybe Epic runs on your own infrastructure on the cloud or on prem, um, as an option. And that's kinda what their stuff hanging it's kind of alluding to in his quote. I mean I once, I'm not really surprised cause I haven't seen a lot of high trust HIPAA certification conversation coming from the Google. Uh, but you know, they did have the Mayo clinic that they announced and that was a pretty big deal. Uh, recently they did mention some HIPAA compatibility things they had there. So will interesting to see if this, you know, changes in the future. I think it's just interesting that they decided to announce it so loudly. Yeah, that's all interesting too. I don't, I don't know exactly why that was the case unless, you know, and that was going to be a bit of a hit job by Asia or AWS saying, Hey, you're not going to go at Google. You should, uh, you know, leak that to CNBC. Kind of hard. A lot of it, I dunno, it's a little weird. Perhaps the medical, I didn't go with JCP, it's more about their clinical research side of the business rather than, I mean, obviously they have uh, uh, hospitals and, and patient care, but they also did a lot of research and think, um, machine learning on the, uh, the GCP cloud for them. Maybe the best choice for the research side of the business, but maybe not for the patient care side. Yeah, it could be. Speaker 2 31:35 Most large organizations run six or even more monitoring tools. Each of them uses a mixture of data collection techniques from technology providers, open source communities or custom integrations and maintaining dozens of integrations across these tools can be a significant investment. Blue Madora introduces bind plane, not another monitoring platform, but the industry's first monitoring integration as a service by playing could gather data from over 150 technology data sources spanning your entire organization. Remove or reduce your reliance on unexpensive monitoring and SIM solutions by sending log data to Google Stackdriver new Relic or as your monitor, check out the extensive list of integrations all provided at no additional cost. Learn more and sign up for a free trial by visiting blue madora.com/cloud pod. The links available in our show notes and as a bonus for cloud pod listeners, blue Madora are offering Google compute platform credits to help get you started by and plain seamlessly stream hybrid cloud and on premise metric and log data Speaker 1 32:41 well onto a set container security topic near and dear to Google's heart. If you're serious about security and Kubernetes, the CIS Kubernetes benchmark will help you a ton. The new 1.5 version of the K benchmark. It was released recently to cover, uh, up to Kubernetes one.one five and these are recommend, Oh, so then a Google also has released a child to the CIS K Kubernetes benchmark that is a child, uh, upstream version that'll allow you to provide a GK hardening guidelines to also help meet those CIS guidelines. So if you're using Kubernetes on any platform, you can take a look at the open source, uh, version of the CIS and if you're using GKE, there's a specific guidelines and recommendations to you to make your system perfect. And those will be found in the security health analytics dashboard of Google cloud. Speaker 2 33:27 I'm going to be really nitpicky now and I'm going to say, wouldn't it be way more sensible to have the version of the benchmark duck match version of Cuban SES? Speaker 1 33:38 You would think so. I mean it'd be nice if they matched cause then it'd be really easy to know which version you're dealing with. Speaker 2 33:43 Please. Yeah, that'd be on that. That's pretty good. I think more and more like the cloud providers need to start telling their customers how to use the tools that they're built properly because there's so much guesswork happens and people try and reinvent more people reinvent the same, uh, the same things over and over again. And documentation like this will, will definitely save time and being an enabler. Um, more importantly I think is to tell the auditors what they should be looking for so that it makes it easy for auditors to look notes, built the best practice and give it a path. Yeah, we mentioned that before and as a, it's um, I love the idea of telling you what they said, what they should be looking for. They don't look at, Oh, don't look over here. Don't go here. Speaker 1 34:29 They really, yeah, they really don't know. It's all new to them. I mean, I've had to explain basic CICT practices live in industry standards for, you know, 10, 15 years now to, you know, to auditors. They just don't understand it. They aren't used to a very old legacy way of doing software development. Actually the this security health analytics dashboard that is in GCP is really impressive. Um, you know, we're working on a small project or on GCP and potentially looking at doing some multi-cloud stuff with GCP at the day job and the person who's driving that effort every day, he sends out a snapshot of the security health analytics dashboard, how they've resolved these open security items as they're trying to get the security team on board. And the kind of, it's nice for the marketing of that, like, Hey see you each week. Each day we're reducing and you know, we have something similar for AWS. We can start it, but we had to buy it from, you know, companies like Evan and IO to get that same capability. That's really nice. That's just kind of built into Google. I am impressed with some of those technologies that are just there now that I had to build or implement, you know, four or five years ago on AWS. Speaker 2 35:25 The same capability. It's nice to be able to turn some of those things off though. You know, I know that the sky is falling, the sky is falling. This ports open to the public. Well, yeah, it's meant to be. Speaker 1 35:35 That's how I feel about the public cloud or the public three bucket warning right now. It's, you know, they've got, they made it really difficult to do that now, but like I really want this one to be public because it's a website and it takes a lot more clicks, a lot more effort now to make that, which is good. And also kind of like, well, if I know what I'm doing and let me, let me bypass some of this. It's just so policing she used to do for the CLI. You don't have to type in. Yes, I really do want to make this bucket public. Please scan the passports of all eight great-grandparents, that kind of thing. As I said, a little excessive. If you are working on the enthos, uh, Google has a new masterclass and hybrid cloud architecture and management for you. The new architecting hybrid cloud infrastructure with Anthem teaches you how to modernize, observe, secure and manage your applications using Istio powered service mesh. Speaker 1 36:22 And Kubernetes, whether you're on premises, on Google cloud or distributed across both or even in a competitive cloud. The courses, this is a mix of lectures and hands on labs. You learn about compute networking, service mesh, configured management and their underlying control plane so you can understand the full scope of the platform capabilities. The courses delivered in three parts, the first being the hybrid cloud infastructure foundations with <inaudible>. The next is the hybrid cloud service mesh and then the last is the hybrid cloud multi cluster with anthros. So if you are looking at Anthem looking at hybrid cloud and you are looking for a class on this, this is now available to you from Google, either in person or on a self paced online through quick labs. Just don't forget not to spit up the anthers service less. You want to spend 10 K a month, a minimum minimum for a year. Speaker 1 37:10 That's quite a pricey. I mean surely they must have seen my laces and things like that if they get off of training. Yeah, I just went through a quick labs, those type of things. It's all, it's all virtualized and available to through the lab. But yeah, if after that you're like, Oh, I'm gonna turn that on and just know you ain't $120,000 commitment in spend. So just be, be aware before you look like in the gooey or in the CLI command and make that mistake. And then a Google cloud on the other side of the bad news from Epic has announced positive news and that they have grown the travel industry presence with the saber and Lou Fonza deals. So they apparently have signed a 10 year agreement with saber Corp. Uh, we'll see Google become the preferred cloud provider. Uh, saber operates a payment platform that processes more than 260 billion in travel related purchases annually for companies such as airlines and hotel chains. Speaker 1 37:58 And it posted revenues of 3.7 8 billion for its most recent fiscal year. So that's a very large customer. It's very embedded and very ingrained in the world of travel. Uh, if you've ever watched your flight attendant at the gate, you know, you need to move your seat or change your flight. And she's frantically typing away at your keyboard. Most likely she is in a Sabre system doing that work for you on the fly. They have a quote here from the chief executive officer of sabers and Sean and Minky as our preferred cloud provider and broader strategic partner, Google cloud will help to accelerate our digital transformation ability to create new marketplace and critical products and systems focused on our customer needs for decades to come. And then the other part of the announcement was the Fonza, uh, which is a German, uh, Germany's flag carrier for airlines and the largest aviation group globally with close to 40 billion of revenues of 2018 has also a intensive build, an AI powered flight optimization platform on the Google cloud. Speaker 1 38:47 Though analyze factors such as departure dates and playmate and schedules to identify ways of cutting wait times for fliers. So using an AIML experience from Google to help them in a big way. Now this is a really great announcement. Um, I do wonder how much of this Sabre, uh, announcement is tied to the fact that Google owns the largest, uh, plane tracking system in the world that Sabre has to integrate with. And, uh, if I go in with their cloud, there are some strategic synergies that may be optimized in that scenario. But, uh, overall I think it's a, a nice win for both the Google and the Fonza and Sabre probably not any preferred provider. They're not, not, don't sell provider. That's, that's true. It does not say soul. Yeah. Tell you 10 years as a provider that's pretty nascent. It must be worth something. Speaker 1 39:30 Blah, blah, blah. How much, you know, it could, it would have been easy to say how much instead they quote the size of the business and the revenue. Yup. Yeah, it's a, it's interesting cause, uh, you know, Google, Google bought ITA for 700 million, uh, back in 2010 and now that sound turned into a very large cyber deal for a very large chunk of money to run Sabre on Google cloud. So we don't know that. I mean, we don't know that, but I'm purely speculating that there might be some of the synergies there. Well, no, I just met. We don't know how if it is going to be significant spend. That's true. Well, good. Well happy to see Google Hailes one when this week after that brutal beating a turf from Epic, uh, which was pretty Epic in its own way. But let's move on to the lightning round. Peter AWS security hub releases the ability to disable specific compliance controls. Didn't just mention this job I did. It's like you read the show notes. There's, I'll see a leak. Speaker 1 40:37 AWS security hub also releases integrations with four new partners. None of the partners that I want to work with, so I appreciate it. None of this other than Slack. Slack is one of those. Nothing says security than a molten by posting events to Slack, I suppose. Yeah. Arizona Aurora supports the read, committed isolation level on read replica. Guys, I'm glad to know about this now because I've committed it to memory. Ooh. Ah, he's, he's trying to game the system pacer. He knows. I like dumb puns. That sounds like Chinese food or some kind of, Hey devil, you eyes elastic Beanstalk, command line interface EBC ILI is now open source. Okay. You went with a public get hub roadmap. You put this on the slide, we talked about it last week and then you literally just moved it to the right and released it. Come on like, you know, don't make your, don't make your public roadmap look better than it actually is. Speaker 1 41:36 You're right. It was imminent to ship. That was a rig that sounded bagging in the past. It is sandbagging AWS glue adds new transforms, purge transition and merge for Apache spark applications to work with data sets and Amazon S three that is superglue. Ooh, Ooh, nice to be as client VPN now supports poured configuration. Who wants this feature and what? What, what purpose? It's only what I like when you first read this, we're like, well maybe this is security through obscurity. Play. Like you're like, yeah, you know, you know you don't wanna use four for three, you wanna use my report. It's more difficult. It only supports one other port, which is one, one nine four. What's the point? Open VPN 1194 was the default port for Oakland VPN and I think did they add <inaudible> or do they add four four three they added 1194 that's the default port for every open VPN client out of the box. Speaker 1 42:28 It's still weird to me. Either way that you know you're, you're announced now supports poor configuration when really you only support one port additional. I kinda wanted, I mean yeah they added 1194 which is today full anyway. But I guess if you've got some outbound proxy which forces that traffic through the proxy, then you definitely need to use it from pull it because it's just not going to work. Hi Beth, somebody asked for it. AWS systems manager now provides flexible reboot options for patching. Hopefully it supports those maintenance window restrictions and blackouts. They announced three weeks ago. Query volume metrics now available for Amazon rod 53 resolver and points should have been day one feature. It's all I can say AWS code pipeline enables stopping pipeline. I think executions. Oh thank God. I can now finally stop my terrible code builds cause I, I just thought, Oh, I'll just push it to the repo, no big deal. And then kicked off a tire pipeline that I can't stop for my 24 hour build. Very slow train wreck. Just unfolding. Speaker 1 43:31 Like I know it's going to fail. It's up 345 I'd have thousands of steps I had to wait for, no, there's a new Azure blueprint for CIS benchmark. Whoo. Yeah. See I, it's tough to get excited about CIS. I mean it's, it's a good standard. I like CIS, but yeah, it's hard to, hard to get. Maybe we should just kill all I stories from now on unless it's interesting and I'm committed to giving the winner to committed. Yeah. Nice. Yes. So again, I listened to episode 55 this morning on the way into the office and I heard the word play thing and I was like, yes, I must hit the word play. Nice. I especially like how you capitalized read, committed in the, uh, in the show notes, CSA to pay to would emphasize that. Just to bring your point home. I mean like if you click the link, yeah. Speaker 1 44:22 It will tell you that Amazon marketing gave us that emphasis, not me. Yeah. Yes. So it's like no cheat. I was not cheating. That was fair and square. That's like New York time to style guide type stuff. If it's a secretly must be capitalized. Well thank you for the points. I now only halfway behind Jonathan. It's still a still a head Jonathan. Yeah. So LA last week, uh, Peter and put a call to action at this part of the show though. Cause I had a bad transition last time and said, uh, you know, what are you guys, uh, looking forward to in the next week or so and we didn't have answers and Peter said that's on you and I'm going to put it, I'm gonna put it back out to you again. What's going on guys? What? I ain't looking forward to anything in the next week. I'm going on a little trip. Speaker 1 45:03 I'm might be able to have an opportunity to check out that Osaka, uh, date offender. I'll be kind of close in. Okayness if you go on WikiLeaks and get the address so you can go check it out. There you go. I actually am going to be a single dad between now and next week cause my wife is escaping on a PFF girls trip and so, uh, she is leaving me with my children and I will be working from home. So by the time we record next week, I will be completely frazzled, so that'll be enjoyable for all of you as we record sometime next week. Uh, and I know Peter, you'll be out. So we will have another amazing guest, uh, to join us here on the show. So I'm looking, I'm looking forward to surviving the next five days without killing my children or having them kill me. Good luck. Speaker 1 45:46 How about, how about you drawn to them? Are you looking forward to, wow, the office is going to be pretty nice without you, uh, staring out the window. I guess you should talk. You should talk to your boss about your new seating location if you want to be stared at. So we're going to be a, we're going to be doing a musical chairs in the office. So again, I again, I'm a times have moved, moved desks in this, this company. It's just, it's just, yeah, I'm up to about nine or 10 moves now. I would like to keep you on your toes. That's how we do it. I at least you don't have like piles and piles of junk like one of our coworkers doesn't just ask. So yeah, I always figured that I keep an empty cargo box under the desk and minimal things so that if I get walked out, I don't have to come back. Nice. Very positive outlook on the unemployment since that's not, that's not like you walked out. Well I believe Jonathan and I have a maintenance window to go rush off to uh, to go resize a database. We're going to do that fun. Uh, Peter, have a great vacation and we will see you on the other side of that and we'll look forward to hearing all about the amazing Japanese whiskey you're going to drink. That's going to be good. I bet it will be. Speaker 1 46:49 And that is the weekend cloud. We'd like to thank our sponsor, Foghorn consulting and blue Madora. Check out our website, the home of the cloud pod where you can join our newsletter, Slack team, send feedback or ask [email protected] or tweet us at hashtag <inaudible> Speaker 3 47:01 <inaudible>.

Other Episodes

Episode 0

February 12, 2020 00:53:08
Episode Cover

The Cloud Pod Faster on Azure... No Wait AWS - Episode 58

Your hosts are joined again by Ryan Lucas (@ryron01) who is filling in for Peter as we recap the week in cloud. A big...

Listen

Episode 0

February 23, 2020 00:38:36
Episode Cover

TCP Talks: Finops in the cloud with Rob Martin - Bonus Ep 2

The most terrifying part of moving to the cloud isn't security, migration techniques or learning new infrastructure as code tools, it is managing that...

Listen

Episode 159

April 07, 2022 00:34:13
Episode Cover

159: The Cloud Pod Suspends Its (GCP) Hosts

On The Cloud Pod this week, Ryan is in the doghouse and he’s been suspended (with full pay). Plus, we’re comfortably numb with AWS...

Listen