AWS re:Invent 2019 is here, and that means 5 days of wall-to-wall releases. But the biggest releases are reserved for the keynotes on Tuesday, Wednesday, and Thursday. And that’s why we’re here. We’ll kick off this post on Tuesday with Andy Jassy’s keynote, then add to it for Wednesday’s Global Partner Summit and Thursday’s keynote from Werner Vogels.
It’s your first-look response to the releases and how they could impact cloud financial management.
AWS CEO Andy Jassy’s presentation is famous for a couple big things: song excerpts from the AWS re:Invent House Band, and establishing a clear theme that gives us a map of where AWS expects innovation to go in the next year.
This year, it was all about transformation and helping you build that next great innovation. He started with establishing the state of the cloud market today, and one point in particular caught our attention.
If you spend a lot of time in cloud, it’s easy to think that everyone’s all-in on cloud. But 97% of companies — many of them some of the largest enterprises in the world — that have yet to migrate to the cloud. Many of the releases will help make the migration easier, while others are going to make it even easier for cloud-first companies.
Whoever the audience, there’s definitely a lot of potential for transformation.
Compute is the single largest cost center on AWS — with good reason. So every re:Invent, AWS announces new compute offerings, and this year is no exception.
Last year, AWS released the first instance running on an AWS-designed chip. This year they build on that success with the release of M6g, R6g, and C6g instances, all powered by AWS-designed Graviton processors. M6g have been added to the general purpose EC2 families today, with R6g and C6g coming soon.
Training machine learning takes a lot of time, effort, and investment, but a large part of the expense from machine learning comes from actually running the system and processing inferences. The new Inf1 instances are designed with AWS Inferentia chips for inferences. They promise up to 40% lower cost per inference, which could have a big effect on ML-heavy infrastructures.
Starting today, Fargate also offers support for EKS. For EKS container systems, this is a big deal, especially since Fargate can be covered by the new Compute Savings Plans. This could make it easier and cheaper to run your EKS infrastructure. It definitely caught the attention of our container engineers, and it’ll be interesting to see the effect it has over the next few months.
After compute, the next biggest cost center is data, which really isn’t a big surprise. In the eternal quest to lower costs and improve efficiency, Jassy also announced several releases in this realm.
When you have many applications sharing the same S3 bucket, controlling access can get complicated. Access Points are designed so you can configure permission to only access specific parts of the bucket. It should make security more scalable with shared data sets, which can lead to increased efficiency and lower costs.
New RA3 instances are designed to allow you to separately scale your compute power and your storage instead of being limited by the local storage on Redshift instances. Designed to work with managed data, these are perfect for performance-intensive workloads.
Amazon Redshift Advanced Query Accelerator (AQUA) uses a hardware-accelerated cache to run Redshit up to 10x faster than any other data warehouse. As Jassy pointed out, this means you can actually do compute on the raw data. That also means a lot more data transfer, so it’ll be interesting to see the effect this has on cloud costs over the next year. For example, there may be an increase in data transfer fees, but lower relative compute times, leading to lower costs overall. Rest assured, we’ll be watching as it rolls out.
Available today in preview, Ultrawarm does to Elasticsearch what Intelligent Tiering did for S3. In essence, it transfers infrequently-accessed data to cheaper storage on the block level. We’ve seen a large adoption of Intelligent Tiering since it’s introduction, largely because it’s a lot of savings for very little lift. Ultrawarm has the potential to have a similarly high adoption rate.
If you’ve ever chosen DynamoDB over Cassandra because AWS DynamoDB is easier to manage, then this is going to be a big release for you. Adding management to Cassandra makes it that much easier to run it on AWS.
Machine learning was a big focus of Jassy’s presentation, with the bulk of the releases coming in this area. Some enabled machine learning, while others utilized machine learning for new services.
Sagemaker got a whole host of new service releases under the umbrella of Amazon SageMaker Studio. The goal is to make it easy for organizations to kick off machine learning efforts. SageMaker Studio includes releases like:
Fraud Detector allows companies to use machine learning to detect fraud without having to build their own systems. We’ve seen machine learning be an incredibly effective way to combat fraud in the past, so this could be a transformative step forward for many companies.
This release probably snagged the attention of every engineer watching the keynote. CodeGuru utilizes machine learning to streamline the code review process, find errors like violation of AWS best practices, concurrency issues, incorrect handling, input validation, and more. Pulling from Amazon code data and 10,000 open source projects, the model has the potential to change how cloud is coded. It certainly piqued the interest of our engineers, and it’ll be interesting to see if the benefits outweigh the service costs. Fortunately, FinOps and TBM can help figure things like that out.
If you use AWS Connect, then this machine-learning-based analysis engine could be a big deal. Contact Lens uses machine learning for contact center analytics to spot problem areas, highlight good interactions, and help improve the quality of your service.
In addition to caring for external customers, companies have to support their internal customers via intranets and internal documentation. Amazon Kendra utilizes machine learning to help manage those internal documents, making it easier to find them using human language questions. The value of this might be hard to quantify in gained hours of work or less IT labor, but if it works as promised, then the lack of stress and increased productivity could be more than worth it.
Compute, storage, databases, and machine learning would be enough for most companies, but not AWS. Jassy also highlighted some big infrastructure changes that could be a game changer for the right use cases.
First announced at last year’s re:Invent, Outposts are finally here. These on-prem versions of AWS infrastructure are designed to make it easy to run a unified hybrid AWS infrastructure. This is going to be a big deal for companies migrating to the cloud, since they can transition to on-prem AWS systems, then do a straight lift-and-shift.
One of the biggest use cases for on-prem data centers is when you need incredibly fast access, such as with video game companies. AWS Local Zones are hyper-local data centers. Positioned close to large population centers, these are perfect for low-latency applications, but selecting one will come at a different cost to other regions or availability zones. The one in Los Angles is currently open to invitation, so we’re not sure how the costs will roll out in comparison to using other Availability Zones.
Developed in partnership with Verizon, AWS Wavelength provides low latency specifically for 5G and mobile applications. Integrated directly into the 5G network, Wavelength can deliver single-digit millisecond latencies to mobile end users. Is it worth it? Well, this is a perfect job for FinOps. If you can get down to the Unit Economics of each user (or user action), then you can see the exact business value of utilizing Wavelength.
AWS is expanding their services to make it easier for the 97% of companies on-prem to migrate to the cloud, while also enabling companies currently on the cloud to do more. Jassy’s presentation brought a lot more options to the table to help your business better accomplish their goals. All you need to do is evaluate the business value of utilizing those services, which is where FinOps and TBM are crucial.
Building or migrating to a cloud infrastructure is a complicated, multi-year project with a lot of moving parts. Fortunately, you’re not alone. There are hundreds of AWS Partner Network (APN) members out there to help you solve your cloud problems, including Apptio Cloudability. The annual Global Partner Summit at re:Invent is a keynote devoted to APN — both for partners and customers.
This is especially important in light of the growing number of enterprises starting their migration journey. With hundreds of thousands of programs, the sheer magnitude of the project makes it daunting, but the experts in the APN can help companies navigate the process.
Helmed by Doug Yeum, Head of Worldwide Channels & Alliances at AWS, the Global Partner Summit highlighted some of these partners with videos and firsthand stories from BP and AvisBudget. Yeum and team also announced releases designed to make it easier for companies to connect to the right partners and get projects moving as cleanly and quickly as possible.
AWS prides itself on providing a wide variety of products and services so companies have the exact right tool. Couple that with growing AWS adoption, and you’ll get an ever-increasing field of partner opportunities.
This is great news for AWS partners and customers alike. Instead of having to settle with tools or partners that do an okay job at a large number of things, they can find partners who specialize in their exact problem. Even better, the space is mature enough that these partners will have plenty of experience, giving you and your team peace of mind.
In fact, during a fireside chat with Andy Jassy, Yeum brought up one of the biggest questions companies ask when evaluating potential partners: “Are there a lot of people who know the technology and have they done the job before?”
To help with this, AWS has core competencies spanning several different fields and practices. For example, Apptio Cloudability was a launch partner in the Cloud Management Tool Cost Optimization competency.
Today, they added to that list by announcing a couple new competencies.
Being able to prepare for, respond to, and recover from emergencies or disasters is essential, but building out those systems requires a very specialized skillset. The Public Safety & Disaster Response Competency identifies partners with experience in these areas.
Retail presents unique problems, including a high number of micro-transactions, security, low latency, and more. The new Retail Competency highlights the partners who specialize in retail solutions, something that will be especially needed as retail enterprises begin their migration journey.
The great thing about competencies is that they allow companies to build multi-partner ecosystems that incorporate best-in-breed experts across every necessary competency. For example, a retailer could work with a partner in the retail competency, then couple it with Apptio Cloudability for cloud financial management and TwistLock for container security.
If you’ve ever used the AWS Marketplace, you’ve probably been shocked by how quickly you can integrate solutions. We’ve actually had companies integrate Apptio Cloudability through AWS Marketplace in a couple hours, then find tens of thousands in savings within the first day.
Several of the new releases make it easier for new companies to get up to the standards of the APN, then streamline the AWS Marketplace supply chain to connect partners and buyers even faster.
Since the beginning, AWS has been all about helping startups just as much as enterprise customers. Jassy even mentioned that in his original internal AWS vision document, he wanted individuals starting new companies to be able to access the same infrastructure as the largest businesses in the world.
The Global Startup Program continues that commitment. Invite-only for mid- to late-startups, the program is designed so help startups accelerate up to join APN and be able to serve enterprise customers.
Negotiated procurement allows companies to negotiate pricing, license terms, payment schedules, and more. With Seller Private Offers, companies and partners can negotiate private offers directly through AWS Marketplace. Once all the terms are agreed upon, AWS Marketplace will take care of provisioning, metering, and integrating software charges into the AWS bill, allowing companies to get the best of both worlds.
Private Offers have been around for a bit in select release, but this GA release also comes with a fee reduction, making it that much easier for sellers to embrace the feature. Hopefully this helps shorten procurement processes, allow companies to take advantage of solutions faster.
Some partners also work with customers to procure other tools as part of their larger project. The new Discovery APIallows select partners (currently only Deloitte, an Apptio Cloudability partner, and DLT) to curate selections — including with data products.
The last thing any of us want is to purchase an AWS tool only to find out it’s incompatible with key services in our infrastructure. AWS Service Ready solves this problem by identifying and validating products from partners that integrate with specific products. Currently available for Redshift and PrivateLink, Service Ready will soon be available for Lambda, RDS, and Outposts.
AWS is only going to continue to add more services, larger footprints, and more capabilities. And the more they grow, the more return customers are going to get on investment in the right partner with the right capabilities. Sure, you’ll always be able to jump into AWS and good results on your own, but the right partners will help you do more and innovate more with your cloud.
In the past, Amazon CTO Werner Vogels’ keynote was packed with releases — sometimes too many to really process. This year, Vogels switched it up, focusing on the technology behind AWS, as well as the engineering benefits of some of their services.
It was definitely a talk for the AWS builders out there, with a lot of the content skewing technical. But the more technical details brought some key points important to best practices for cloud financial management.
If you’ve looked at the specs for EC2 instances in the last couple years, you’ve probably seen the phrase “Nitro hypervisor” thrown around. But is it really the big deal the documentation says it is? Vogels dug into how it works a little more, and yes. Yes, it is.
In essence, the Nitro hypervisor changes how virtualization works with the servers behind EC2 instances. Since Nitro was introduced, latency has dropped, noise as decreased sharply, and throughput has increased. It’s enabled the rapid release of new EC2 instances, as well as providing the backbone behind capabilities like live updates, VMWare, bare metal, and AWS Outposts. If you want to find out more, check out the AWS documentation.
If you apply these lessons to cloud financial management, it explains the technical background behind the best practice of rightsizing EC2 instances to the latest generation. The M4 family, for example, isn’t run with Nitro hypervisor, while the M5 family is. M5 instances run faster, have a higher ECU, and cost less than M4 instances. Switching out M4 usage for M5 instances can often be a quick way to rightsize your infrastructure so you lower your costs without impacting the amount of resources you get.
That being said, it’s not always as simple as replacing all instance usage with the latest generation. Variables like purchased RIs or the expense of refactoring might present more hurdles than it’s worth. But upgrading to Nitro-powered instances is definitely an option that should be explored, and features like Apptio Cloudability’s Rightsizing and RI Portfolio can help you build strategy that will optimize costs without impacting performance.
The open source project Firecracker for containers was released at re:Invent 2018. Over the last year, it’s left its mark on containers and Fargate. AWS Principal engineer Clare Liguori took the stage during the keynote to walk us through exactly what that meant. She highlighted advancements with Firecracker and the effectiveness of Fargate at minimizing container over- and under-provisioning.
Depending on your situation, that more efficient provisioning could give you a financial reason to switch to Fargate. But there are a lot of variables in play, and you want to weigh all of them, including the cost of refactoring, potential bottlenecks if you only have a few engineers who can run containers without Fargate, scale, and more. A solid FinOps practice and tool like Apptio Cloudability can help you make that decision with confidence.
As a follow-up, Vogels presenting an interest piece of data from AWS. Originally, they had expected the biggest container adopters to be smaller start-ups. Instead, they found that enterprise customers were the top users of containers and microservices. He gave a few examples and brought a couple speakers on stage to drive the point home.
And once you dig into that concept a little more, the logic starts to make sense. A big advantage of containerization is the ability to rapidly scale up and down at massive levels, something enterprise infrastructures deal with all the time. Additionally, enterprises are working on a scale that would allow them to take advantage of tiny gains in efficiency that add up over millions of containers. On the flip side, labor costs and additional expense for services like Fargate might offset any savings.
The key is to ensure you have solid cost and usage reporting coupled with a complete overview of costs, including labor, refactoring, RI/Savings Plan coverage, and more.
The vast majority of Vogels’ keynote was spent on lessons, examples, and generally dropping AWS knowledge. So it follows that the one release Vogel announced was all about making knowledge easier to access for engineers, architects, and other builders.
The Amazon Builders’ Library is a learning center filled with articles and best practices from the people who know AWS cloud the best: AWS engineers. At the time of launch, there are 13 articles from 200 to 400 levels covering topics like Architecture and Software Delivery and Operations.
It’ll be interesting to see the library grow over the coming months, and we’ll definitely be watching for any insights that could help with cloud financial management or improved efficiency. Hopefully, this proves to be a valuable knowledge base to help engineers better accomplish their goals!
While Vogels’ keynote was a lot more technical than Jassy’s or the Global Partner Summit, there was still a core financial management lesson that could be gleaned from the presentation: Make sure you account for all of the variables when building financial strategies — and the means more than just costs.
FinOps success comes from increasing efficiency for cost and usage so you can improve the business value of your cloud and better accomplish your goals. Taking advantage of technical innovations and the latest improvements from AWS can be a big part of building those strategies.
AWS’ goal is to make sure that your company has every tool possible to make the most out of their cloud. This mean hundreds of services that can get increasingly more complicated to manage.
But don’t worry. Apptio Cloudability is designed from the ground up to make cloud and hybrid financial management easy, so you can optimize costs and fuel innovation.
Reach out today to get a demo and see for yourself!