A few weeks ago I attended Amazon Web Service (AWS) re:Invent 2014 in Las Vegas for a few days. For those of you who have not yet attended this event, I recommend adding it to your agenda. If you have interest in compute servers, networking, storage, development tools or management of cloud (public, private, hybrid), virtualization and related topic themes, you should check out AWS re:invent.
AWS made several announcements at re:invent including many around development tools, compute and data storage services. One of those to keep an eye on is cloud based Aurora relational database service that complement existing RDS tools. Aurora is positioned as an alternative to traditional SQL based transactional databases commonly found in enterprise environments (e.g. SQL Server among others).
Using the AWS Management Portal for vCenter adds a plug-in within your VMware vCenter to manage your AWS infrastructure. The vCenter for AWS plug-in includes support for AWS EC2 and Virtual Machine (VM) import to migrate your VMware VMs to AWS EC2, create VPC (Virtual Private Clouds) along with subnet’s. There is no cost for the plug-in, you simply pay for the underlying AWS resources consumed (e.g. EC2, EBS, S3). Learn more about AWS Management Portal for vCenter here, and download the OVA plug-in for vCenter here.
November 12, 2014 (Day 1) Keynote (highlight video, full keynote). This is the session where AWS SVP Andy Jassy made several announcements including Aurora relational database that complements existing RDS (Relational Data Services). In addition to Andy, the key-note sessions also included various special guests ranging from AWS customers, partners and internal people in support of the various initiatives and announcements.
Announcements and enhancements made by AWS during re:Invent include:
Hardware security module (HSM) based key managed service for creating and control of encryption keys to protect security of digital assets and their keys. Integration with AWS EBS and others services including S3 and Redshift along with CloudTrail logs for regulatory, compliance and management. Learn more about AWS KMS here
For those who are not familiar, AWS has a suite of database related services including SQL and no SQL based, simple to transactional to Petabyte (PB) scale data warehouses for big data and analytics. AWS offers the Relational Database Service (RDS) which is a suite of different database types, instances and services. RDS instance and types include SimpleDB, MySQL, Postgress, Oracle, SQL Server and the new AWS Aurora offering (read more below). Other little data database and big data repository related offerings include DynamoDB (a non-SQL database), ElasticCache (in memory cache repository) and Redshift (large-scale data warehouse and big data repository).
In addition to database services offered by AWS, you can also combine various AWS resources including EC2 compute, EBS and other storage offerings to create your own solution. For example there are various Amazon Machine Images (AMI’s) or pre-built operating systems and database tools available with EC2 as well as via the AWS Marketplace , such as MongoDB and Couchbase among others. For those not familiar with MongoDB, Couchbase, Cassandra, Riak along with other non SQL or alternative databases and key value repositories, check out Seven Databases in Seven Weeks in my book review of it here.
Aurora is a new relational database offering part of the AWS RDS suite of services. Positioned as an alternative to commercial high-end database, Aurora is a cost-effective database engine compatible with MySQL. AWS is claiming 5x better performance than standard MySQL with Aurora while being resilient and durable. Learn more about Aurora which will be available in early 2015 and its current preview here.
AWS will be adding a new C4 instance as a next generation of EC2 compute instance based on Intel Xeon E5-2666 v3 (Haswell) processors. The Intel Xeon E5-2666 v3 processors run at a clock speed of 2.9 GHz providing the highest level of EC2 performance. AWS is targeting traditional High Performance Computing (HPC) along with other compute intensive workloads including analytics, gaming, and transcoding among others. Learn more AWS EC2 instances here, and view this Server and StorageIO EC2, EBS and associated AWS primer here.
Containers such as those via Docker have become popular to support developers rapidly build as well as deploy scalable applications. AWS has added a new feature called EC2 Container Service that supports Docker using simple API’s. In addition to supporting Docker, EC2 Container Service is a high performance scalable container management service for distributed applications deployed on a cluster of EC2 instances. Similar to other EC2 services, EC2 Container Service leverages security groups, EBS volumes and Identity Access Management (IAM) roles along with scheduling placement of containers to meet your needs. Note that AWS is not alone in adding container and docker support with Microsoft Azure also having recently made some announcements, learn more about Azure and Docker here. Learn more about EC2 container service here and more about Docker here.
In addition to announcing new higher performance Elastic Cloud Compute (EC2) compute instances along with container service, another new service is AWS Lambda. Lambda is a service that automatically and quickly runs your applications code in response to events, activities, or other triggers. In addition to running your code, Lambda service is billed in 100 millisecond increments along with corresponding memory use vs. standard EC2 per hour billing. What this means is that instead of paying for an hour of time for your code to run, you can choose to use the Lambda service with more fine-grained consumption billing.
Various application code deployment models
Lambda service is a pay for what you consume, charges are based on the number of requests for your code function (e.g. application), amount of memory and execution time. There is a free tier for Lambda that includes 1 million requests and 400,000 GByte seconds of time per month. A GByte second is the amount of memory (e.g. DRAM vs. storage) consumed during a second. An example is your application is run 100,000 times and runs for 1 second consuming 128MB of memory = 128,000,000MB = 128,000GB seconds. View various pricing models here on the AWS Lambda site that show examples for different memory sizes, times a function runs and run time.
How much memory you select for your application code determines how it can run in the AWS free tier, which is available to both existing and new customers. Lambda fees are based on the total across all of your functions starting with the code when it runs. Note that you could have from one to thousands or more different functions running in Lambda service. As of this time, AWS is showing Lambda pricing as free for the first 1 million requests, and beyond that, $0.20 per 1 million request ($0.0000002 per request) per duration. Duration is from when you code runs until it ends or otherwise terminates rounded up to the nearest 100ms. The Lambda price also depends on the amount of memory you allocated for your code. Once past the 400,000 GByte second per month free tier the fee is $0.00001667 for every GB second used.
Why would you use AWS Lambda vs. provisioning an Container, EC2 instance or running your application code function on a traditional or virtual machine?
View AWS Lambda pricing along with free tier information here.
AWS is increasing the performance and size of General Purpose SSD and Provisioned IOP’s SSD volumes. This means that you can create volumes up to 16TB and 10,000 IOP’s for AWS EBS general-purpose SSD volumes. For EBS Provisioned IOP’s SSD volumes you can create up to 16TB for 20,000 IOP’s. General-purpose SSD volumes deliver a maximum throughput (bandwidth) of 160 MBps and Provisioned IOP SSD volumes have been specified by AWS at 320MBps when attached to EBS optimized instances. Learn more about EBS capabilities here. Verify your IO size and verify AWS sizing information to avoid surprises as all IO sizes are not considered to be the same. Learn more about Provisioned IOP’s, optimized instances, EBS and EC2 fundamentals in this StorageIO AWS primer here.
In addition to compute and storage resource enhancements, AWS has also announced several tools to support application development, configuration along with deployment (life-cycle management). These include tools that AWS uses themselves as part of building and maintaining the AWS platform services.
Management, reporting and monitoring capabilities including Data center infrastructure management (DCIM) for monitoring your AWS resources, configuration (including history), governance, change management and notifications. AWS Config enables similar capabilities to support DCIM, Change Management Database (CMDB), trouble shooting and diagnostics, auditing, resource and configuration analysis among other activities. Learn more about AWS Config here.
AWS announced a new service catalog that will be available in early 2015. This new service capability will enable administrators to create and manage catalogs of approved resources for users to use via their personalized portal. Learn more about AWS service catalog here.
To support code rapid deployment automation for EC2 instances, AWS has released CodeDeploy. CodeDeploy masks complexity associated with deployment when adding new features to your applications while reducing human error-prone operations. As part of the announcement, AWS mentioned that they are using CodeDeploy as part of their own applications development, maintenance, and change-management and deployment operations. While suited for at scale deployments across many instances, CodeDeploy works with as small as a single EC2 instance. Learn more about AWS CodeDeploy here.
For application code management, AWS will be making available in early 2015 a new service called CodeCommit. CodeCommit is a highly scalable secure source control service that host private Git repositories. Supporting standard functionalities of Git, including collaboration, you can store things from source code to binaries while working with your existing tools. Learn more about AWS CodeCommit here.
To support application delivery and release automation along with associated management tools, AWS is making available CodePipeline. CodePipeline is a tool (service) that supports build, checking workflow’s, code staging, testing and release to production including support for 3rd party tool integration. CodePipeline will be available in early 2015, learn more here.
AWS continues to invest as well as re-invest into its environment both adding new feature functionality, as well as expanding the extensibility of those features. This means that AWS like other vendors or service providers adds new check-box features, however they also like some increase the depth extensibility of those capabilities.
Besides adding new features and increasing the extensibility of existing capabilities, AWS is addressing both the data and information infrastructure including compute (server), storage and database, networking along with associated management tools while also adding extra developer tools. Developer tools include life-cycle management supporting code creation, testing, tracking, testing, change management among other management activities.
Another observation is that while AWS continues to promote the public cloud such as those services they offer as the present and future, they are also talking hybrid cloud. Granted you have to listen carefully as you may not simply hear hybrid cloud used like some toss it around, however listen for and look into AWS Virtual Private Cloud (VPC), along with what you can do using various technologies via the AWS marketplace.
AWS is also speaking the language of enterprise and traditional IT from an applications and development to data and information infrastructure perspective while also walking the cloud talk. What this means is that AWS realizes that they need to help existing environments evolve and make the transition to the cloud which means speaking their language vs. converting them to cloud conversations to then be able to migrate them to the cloud. These steps should make AWS practical for many enterprise environments looking to make the transition to public and hybrid cloud at their pace, some faster than others. More on these and some related themes in future posts.
The AWS re:Invent event continues to grow year over year, I heard a figure of over 12,000 people however it was not clear if that included exhibiting vendors, AWS people, attendees, analyst, bloggers and media among others. However a simple validation is that the keynotes were in the larger rooms used by events such as EMCworld and VMworld when they hosted in Las Vegas as was the expo space vs. what I saw last year while at re:Invent. Unlike some large events such as VMworld where at best there is a waiting queue or line to get into sessions or hands on lab (HOL), while becoming more crowded, AWS re:Invent is still easy to get in and spend some time using the HOL which is of course powered by AWS meaning you can resume what you started while at re:Invent later. Overall a good event and nice series of enhancements by AWS, looking forward to next years AWS re:Invent.