AWS re:Invent kicked off Monday with certifications, labs and bootcamps before the Expo floor opened in the afternoon. And before you knew it, it was Monday night. That meant it was time for Monday Night Live.
New Amazon EC2 Instances: C5n, A1 and P3dn
C5 and P3 each got a new instance type added to their families with the introduction of the C5n and P3dn (coming soon) instances. AWS also introduced the brand new A family with the release of the A1 instance. Here’s a quick rundown of our first impressions:
- A1: The new A1 instances are custom-built using AWS-designed silicon. They’re specifically made to balance cost and performance for ARM-based workloads, such as ARM CI/CD environments. They may also be a good choice for lower-compute workloads, like web servers or containerized microservices. Our initial tests showed that for average-compute workloads, M5 and T3 will still be cheaper, since they’ll finish the compute in less time than the A1, but we’re interested to see how A1 is used in the future.
- gives them a lower rate that could be perfect for low-compute workloads like websites. Our initial testing with higher-compute
- C5n: C5n joins C5 and C5d as powerful compute instances built on 3.0 Ghz Intel® Xeon® Platinum 8000 series processors, but with one big difference — 100 Gbps of network bandwidth. It’s a big jump from the 25 Gbps of C5 and C5d. Couple that with 192 GiB of RAM, and C5n is a powerhouse of computing.
- P3dn: The upcoming P3dn will have 2x the GPU memory and 1.5x the CPUs as P3, allowing more power per node for high compute applications. For example, machine learning workloads can be distributed across fewer nodes.
Again, these are only our first impressions. It’ll be interesting to see what happens with these new instances in the next few months and how they affect optimizing EC2 costs!
Global Partner Summit Keynote
Tuesday morning kicked off bright and early with the Global Partner Summit Keynote, led by Terry Wise, the AWS VP of Global Alliances and Channels. During the two hour presentation, he was joined by:
- Pebbles Sy-Manalang – CIO, Globe Telecom
- Ramin Sayar – President & CEO, Sumo Logic
- Brad Jackson – CEO, Slalom
- Bernd Heinemann – Board Member, Allianz Germany
- David McCann – VP of AWS Marketplace, Service Catalog & Migration Services
- Andy Jassy – CEO, Amazon Web Services
There were a lot of great insights and data during the keynote, but a few lessons really stood out to us.
1. Partners are essential
More enterprises are embracing the cloud, and many of them are doing it through AWS partners. We’ve seen a similar trend with many of our customers. Trusted partners are helping businesses navigate the hurdles of building their cloud from migration to setting up key cloud processes. (We’d like give a shout out to our partner, Edrans, and congratulate them on their new status as an AWS Premier Consulting Partner. Check out our recent webinar with them!)
Terry also spent time specifically highlighting the AWS Well-Architected Framework. With so many companies looking for guidance as they craft their cloud infrastructure, this framework is essential. We couldn’t agree more, which is why we’re proud to have achieved the AWS competency for Cloud Management Tools. We’ve chosen to focus on the Cost Optimization pillar of the Well-Architected Framework, and are proud to help our partners achieve their cloud goals.
2. Containers & serverless are still growing
In 2018, our State of the Cloud report showed massive growth of containers and serverless technologies. The trend is only continuing. AWS has seen more and more companies, including many enterprises, investing heavily in containers. According to Terry, containers are becoming the de facto standard for new applications and for refactored applications.
To offer greater support of containers, there were a couple key releases:
- New AWS Container Competency Program — The new competency includes Cloudability partners Datadog and Twistlock.
- New AWS Marketplace for Containers — Now AWS users can buy and deploy container products through the Marketplace.
3. AWS partners need to be able to think internationally
Pebbles Sy-Manalang is the CIO of Globe Telecom, based in the Philippines, while Bernd Heinemann is a Board Member of Allianz Germany. Both of them shared compelling stories of how their companies are utilizing AWS. Allianz used the cloud to ensure that every customer has the personal connection that’s so crucial in the insurance industry, while Globe is bringing connectivity to a population spread across thousands of islands.
The truth is, the US is ahead when it comes to enterprise-level adoption of the cloud. AWS CEO Andy Jassy said that we’re in the early stages of mainstream enterprise adoption in the US, and 12-13 months out in the rest of the world. But we’re still seeing massive adoption of the cloud across the globe. It’s why we opened our London and Sydney offices, and why some of our strongest business relationships, including Atlassian, are with companies based outside the US.
The cloud is changing how all of us relate to technology and each other, no matter where we live.
4. A defined cloud culture is vital for companies
Brad Jackson, CEO of Slalom, gave some great insights into the best practices of companies that have successfully embraced the cloud. He specifically called out the importance of getting 100% commitment to the cloud. He also advocated:
- Driving relentlessly to build teams dedicated to creating a strong cloud
- Being very focused on metrics and measuring everything possible
- Taking a very long-term view
We totally agree, and it’s why we advocate building a cloud cost culture within your company. The cloud and various practices like DevOps decentralize IT operations, making it essential for everyone to be on the same page. This won’t happen by accident — it takes a concerted effort by leadership to create the right culture.
In a dovetail to this concept, David McCann announced the new AWS Private Marketplace. Building on AWS Organizations, this allows companies to create curated marketplaces for members of their organizations. If used correctly, it promises to create that ideal blend of freedom and oversight that allows cloud innovation to flourish in a company.
5. Partner insights from Andy Jassy
Terry’s chat with Andy had a bunch of great insights, and we’re going to pull out a few choice ones:
- Enterprises are using cloud migration as an opportunity to re-evaluate long-standing relationships with IT partners.
- Many companies are looking to move away from old guard database providers, and the wish to push their databases to the cloud is driving their migration.
- Part of the help partners can provide is helping companies get on the cloud, but the other large part is helping companies organize their teams.
- Data Lake is a popular buzzword, much like Big Data or Machine Learning in the last couple years, and companies are pushing to the cloud to build their own Data Lakes.
- Enterprises starting the cloud journey need help with end-to-end portfolio analysis and migration planning.
The biggest takeaway is that enterprises are ready to embrace the cloud, but are cautious about how to proceed. It’s our responsibility as cloud innovators to help them get the answers they need.
Challenges of Container Cost Allocation and Chargeback — with Erik Onnen
Containers were also the subject of the talk our CTO, Erik Onnen, gave in the Kumo Theater late Tuesday morning. As we’ve talked about before, containers pose specific challenges when allocating costs. Erik discussed key points of container cost allocation, including:
- Achieving cost visibility, allocation and chargeback with containers
- Accounting for the variable cost basis with multiple algorithms for cost sharing
- Compensating for limited visibility into task resource usage
- Incorporating diverse models for resource allocation
Container use is rapidly increasing, with more and more enterprise companies adopting their use. It’s vital for any company working with containers to create a comprehensive process for correctly allocating container costs and resource use.
AWS Ground Station
We’re honestly including this release just because it involves satellites and space. Before today, accessing one of the thousands of satellites in orbit meant expensive, long-term contracts with companies that own ground stations. Well, AWS is now the proud owner of two ground stations, with another ten planned for 2019. And in typical AWS style, they’re making the use of those ground stations available on a pay-as-you-go basis.
You can find out more details about this new service on the AWS Blog. We’ll be interested to see how much use it gets in the future, but we love seeing AWS launch new services and push the boundaries of cloud computing to the stratosphere — and beyond.
AWS re:Invent was in full swing on Wednesday. We saw tons of booth traffic all day, gave away hundreds of super-soft shirts and delivered plenty of shiny Cloud Control 3rd Edition copies.
But when it comes to announcements and news, all eyes are on the morning’s keynote.
Keynote — Andy Jassy, AWS CEO
The keynotes on Tuesday and Thursday are two hours long, but Andy simply had too much news. His three hour keynote on Wednesday was packed with announcements, insights and plenty of assists from the AWS house band. How packed? Well, all told, the keynote included 20 new announcements.
- Amazon FSx for Windows File Server – Fully-managed Windows file system on Windows native servers
- Amazon FSx for Lustre – Fully-managed file servers for compute-intensive workloads
- Amazon DynamoDB On-Demand – Flexible DynamoDB with no capacity planning necessary
- Amazon Elastic Inference – GPU-Powered Deep Learning Inference Acceleration
- Amazon SageMaker Ground Truth – Build accurate datasets and reduce costs at the same time
- AWS Marketplace for Machine Learning – ML algorithms and model packages are now available in the Marketplace
- Amazon SageMaker RL – Managed Reinforcement Learning with Amazon SageMaker
- Amazon Forecast – Time series using Amazon’s forecast algorithms
- Amazon S3 Glacier Deep Archive – New long-term data archival class for S3
- AWS Control Tower – Automate setting up a well-architected multi-account AWS environment
- Amazon Textract – Optical Character Recognition to extract data from most documents
- Amazon Personalize – Real-time personalization and recommendation
- AWS Outposts – Bringing AWS hardware and software on-premises
- Amazon RDS on VMware – Fully-managed service for on-premises databases
- Amazon Quantum Ledger Database – Fully-managed ledger databases
- AWS Managed Blockchain – Managed blockchain service supporting Hyperledger Fabric and Ethereum
- Amazon Timestream – Database service designed specifically for time-series data
- AWS Lake Formation – Fully-managed service will help you to build, secure and manage a data lake
- AWS Security Hub – Centrally view and manage security alerts and automate compliance checks
- AWS DeepRacer – Go hands-on with Reinforcement Learning with the DeepRacer
AWS Chief Evangelist Jeff Barr did a great job summarizing these releases, so head over to the AWS blog for a full breakdown. But there were some common themes to the keynote and the releases. Looking at these themes gives us an idea of where the industry is going, what problems companies will be looking to solve in the next year — and the areas where companies will need to watch their cloud costs.
Theme #1: More Data Options than Ever
Yesterday in a Q&A during the Partner Summit Keynote, Andy touched on how much the industry is focusing on data, and especially data lakes. During today’s keynote, he followed up on that theme, and brought the releases to support it.
S3 saw two new services introduced with Intelligent Tiering (introduced earlier this week) and Glacier Deep Archive. At the same time, Elastic File System (EFS) saw the introduction of EFS-IA for cheaper long-term storage, while the release of DynamoDB Read/Write Capacity On Demand prevents unnecessary, costly over-provisioning for flexible database loads. The combination of these releases makes it easier than ever to store data in the exact right service, whether it’s S3-IA, Glacier Deep Archive or precisely provisioned DynamoDB. All told, this points to a trend of giving companies more tools to optimize their data storage costs.
Specialized data technology also got a boost during the keynote with the introduction of Timestream, Quantum Ledger Database and Managed Blockchain. Support for these applications shows AWS’ commitment to providing a wide variety specially-designed services, but it also shows that AWS anticipates these technologies seeing wider use in the future.
The release of FSx for Windows and FSx for Lustre also caught our attention, especially when coupled with the release of AWS Transfer for SFTP earlier this week. It turns out that AWS has the largest number of Windows public cloud IaaS instances running, but there are still plenty of systems waiting to migrate, and these releases provide additional tools to make that happen. Many of those older systems require file systems, which is where FSx comes in. And if the systems need FTP to access their data, Transfer for SFTP makes migration possible.
Speaking of making migration possible…
Theme #2: Bridging the Cloud Gap
In the last two days, there’s been a lot of conversation about moving enterprises to the cloud. The process is often very involved, taking 2-5 years in total, according to Andy. The AWS users behind these migrations can be less concerned with innovating new cloud solutions at a granular level and more focused on building clouds that conform to established cloud best practices and systems.
Several launches during Wednesday’s keynote are designed with those users in mind. Control Tower, Lake Formation and Security Hub are all tailor-made to make the enterprise migration process easier. Control Tower helps you set up a well-architected, multi-account AWS environment, Lake Formation helps you create a data lake and Security Hub gives you a single place to centrally manage security.
Andy was also joined by Pat Gelsinger, CEO of VMWare, to reaffirm their partnership and make an announcement of an upcoming product built on that partnership — AWS Outposts. These on-premises servers use AWS hardware (the same that’s in their data centers), and can either run VMWare Cloud on AWS or AWS native applications.
When you put all of this together, it points towards a surge of more companies moving to the cloud in the near future. It also follows many of the same observations we’ve made at Cloudability. For several years, the cloud attracted early adopters and innovators looking to dig into the minutiae and create their own solutions. Now, we’re seeing companies looking for best practices, frameworks and processes that have proven to be successful. Fortunately, those best practices have had time to be developed and tested. We’ve seen similar behavior with cloud cost management, and now are able to point to best practices for creating a culture of cloud cost management in companies.
Theme #3: Machine Learning
In the same keynote, the launches went from support companies migrating older systems and data to the cloud to enabling the application of machine learning, which really shows the breadth of AWS offerings.
Amazon Personalize, Forecast and Textract provided greater application opportunities of machine learning. Personalize (currently in preview) streamlines the process of designing, training and deploying a machine learning model for personalization. Textract (also in preview) uses machine learning to accurately extract data from documents. Forecast uses Amazon’s own years of experience to help users take advantage of machine learning to manage supply chains. These more “ready-made” machine learning options open the door for more companies utilizing machine learning.
At the same time, AWS also enabled more hands-on machine learning efforts with the release of SageMaker Ground Truth, Elastic Inference and SageMarker RL, not to mention adding machine learning algorithms and models to the AWS Marketplace. This combination should enable machine learning teams to build stronger data sets, more efficiently process that data and build out more reinforcement learning models.
All told, AWS is focusing a lot of resources to enabling machine learning across the spectrum of applications. The combination of releases also opens the door for greater optimization efforts with machine learning setups, including cost optimization. Fortunately, some of the more ready-made solutions will free up resources to build the surrounding best practices and processes necessary to build out a cloud cost management culture that can maximize machine learning impact.
AWS DeepRacer & AWS DeepRacer League
Okay, technically this was a keynote machine learning release, but the DeepRacer deserves its own section based on awesomeness alone. A fully autonomous 1/18th scale race car driven by reinforcement learning, the DeepRacer is a great way to experiment with reinforcement learning — and have some fun at the same time. There are some models at re:Invent being taken for a spin by attendees, but you can pre-order one now to get your hands on next year when they’re widely released. (Cloudability has already put our pre-order in!)
And because every race car needs a racing league, AWS also announced the AWS DeepRacer League, set to launch in early 2019. Teams will compete for a chance to race for the AWS DeepRacer Cup at re:Invent 2019. The engineering and data science teams at Cloudability are already making plans for our DeepRacer. See you on the track!
It’s Thursday, and in typical AWS style, we’re finishing re:Invent with flair. (And no, we’re not just talking about the Skrillex-headlined re:Play!) The Thursday morning keynote was filled with great releases, key insights and enough inspiration to fuel innovation through 2019 — and beyond.
Keynote — Dr. Werner Vogels, CTO, Amazon
Werner wasted no time in diving into the technology behind AWS. During the keynote, he went under the hood of AWS’ data services and announced several new serverless releases. He was also joined by:
- Mai-Lan Tomsen Bukovec – VP Amazon S3, AWS
- Ethan Kaplan – CPO, Fender Digital, Fender Musical Instruments
- Holly Mesrobian – Director of Engineering, AWS Lambda, AWS
- Yuri Misnik – Executive General Manager, Business Enabling Technology, National Australia Bank
The combination of use cases and deep looks at key AWS products showed off AWS’ approach to data, their commitment to serverless technology and how they’re helping enterprises innovate.
Let’s look at some of the highlights and insights from Werner’s keynote!
#1: Why and How AWS Protects Data Integrity
Werner kicked off his keynote with a story from before AWS when a code error shut down the entire Amazon.com database for 12 hours. To make matters worse, it was in December, and on the last day to order items with Super Saver Shipping for delivery before Christmas.
After this painful experience, the team worked hard to develop a cell-based architecture. The idea is simple — as Werner said, “Everything fails all the time. What you want to achieve is that you minimize the impact of such a failure.” He calls it reducing your blast radius, and the concept is at the core of AWS’ data storage.
Werner then took a deep dive into how the cell-based approach is utilized with Aurora and DynamoDB, showing how the structure leads to increased efficiency, more durability and reduced latency. Then he invited Mai-Lan Tomsen Bukovec, the VP of S3 at AWS, onto the stage.
Similar to the story Werner told, Mai-Lan told the story of how S3 has grown from eight microservices to more than 235. And yet the core culture of durability remains the same. She dug into how that approach has allowed them to design for eleven nines of durability. As an example, she called out that they don’t look at the mean time to repair problems. Instead, they try to solve and plan for the worst-case time to repair.
Werner then summed up the data section by looking at RedShift, and how it’s grown over just the last six months to be 3.5x faster, with 87% of customers not having significant wait times. And in an illustration of AWS’ customer approach, Werner then announced Amazon Redshift concurrency scaling — designed to speed up Redshift and eliminate wait times for that last 13%.
Overall, the story of AWS’ approach to data was a fascinating look into their process, but is even more interesting when considered with the other keynotes delivered this week. A common thread through all of the keynotes was the robustness of the AWS data services, and how more enterprises are taking advantage of that robustness to make the migration to the cloud, even companies in highly-regulated industries like finance or healthcare.
By highlighting AWS’ strategic approach to data, Werner and Mai-Lan provide the technical backup for AWS’ focus on enterprise data. And to prove it, Werner specifically called out a recent migration where one of the largest Oracle data centers was shut down and the entire workload migrated to Redshift.
All told, the three keynotes have painted a strong picture of the breadth and depth of AWS’ services for storing and utilizing data. All those options give companies of every size the tools they need to protect, store and use their data. There are a lot of moving parts, which speaks to the importance of having a cloud cost management solution to give you the insights you need to make the most out of your data.
#2: Massive Serverless Expansion — And Plenty of New Releases!
Serverless has been another theme of the keynotes. Every speaker mentioned the growth of serverless technology, and it’s no surprise that it was a major part of Werner’s keynote. He called out how customers are flocking to serverless to build stronger applications, faster.
Holly Mesrobian, Director of Engineering for AWS Lambda, then dug deeper into Lambda, which is now processing trillions of requests every month. Specifically, she dug into how AWS provides the “fully-managed” part of the Lambda service, everything from workload optimization for both sandboxes and servers to isolating workloads for security. Her talked helped show the strength of Lambda as a whole, and how it helps free up developers to focus on building apps instead of managing servers.
Which is good, because there’s probably going to be an increase in Lambda use with the new serverless releases Werner announced:
- New AWS Toolkits – Open source toolkits for PyCharm, IntelliJ and VS Code
- Ruby Support for Lambda – Adding to the existing support for 6 other popular languages
- Custom Runtimes – Allowing you to use any Linux compatible language runtime
- Lambda Layers – Upload shared libraries as a Lambda Layer instead of deploying it every time. (Libraries now available from Cloudability partners Datadog and Twistlock!)
- Nested Applications Using Serverless Application Repository – Nested applications can now be used with the AWS Serverless Application Repository
- Step Functions Service Integrations – Connect and coordinate AWS services without writing code
- WebSocket Support for API Gateway – Build real-time two-way communication applications
- ALB Support for Lambda – Use Application Load Balancers with Lambda functions
- Amazon Managed Streaming for Kafka – Fully-managed Apache Kafka service
Combined, the releases fill out Lambda’s offerings across IDEs, Languages, Programming Models and Workflows. As a result, developers have a lot more options for using Lambda, which should fuel even greater adoption of Lambda and serverless technology. It also increases the potential complexity of tackling and allocating Lambda costs and usage.
Which brings us to the last insight from the keynote:
#3: Well-Architected Frameworks are Essential for Best Results
Yuri Misnik from National Australia Bank (NAB) and Ethan Kaplan from Fender Musical Instruments both shared their story of building out cloud ecosystems. From NAB, the process is a migration of applications with varied requirements and regulations, while for Fender the project was creating a new, cloud-based application system.
With the support of AWS, both companies have had success in the cloud. But there was another common thread: both of them followed the best practices behind the AWS Well-Architected Framework. Werner spoke about the importance of ensuring your company is following the Well-Architected Framework, including all five pillars:
- Operational Excellence
- Performance Efficiency
- Cost Optimization (Note: Cost optimization is Cloudability’s specialty!)
To help companies be well-architected, AWS and partners can conduct reviews, which can be incredibly helpful for companies and provide an expert perspective. But a new release allows companies to have a more continuous culture of a well-architected framework: the AWS Well-Architected Tool. This self-service tool allows cloud architects to compare their workloads to the framework and make any necessary improvements.
A Cloud Cost Culture Is Essential for AWS Success
AWS’ releases have expanded their already-expansive portfolio of offerings. Companies have more options all the way from data center migration of older systems to innovative cloud-first applications.
But no matter how you’re using AWS, getting the most out of your cloud means you have to build a culture of cloud cost management. This will help you optimize your costs and help make sure you get the most cloud for your investment.