Mar 14, 2017
Edward Hsu
D2iQ
5 min read
We're excited to announce Mesosphere DC/OS 1.9, which takes production container operations and data services automation to the next level. This release further expands our data services ecosystem, and adds three key new sets of features to make it easier for you to build and deploy modern, data-rich applications quickly and reliably. We are proud to announce the availability of new data services on DC/OS, each now deployable with a single click: Alluxio, Datastax Enterprise, Couchbase Server, Elastic Stack, and Redis Labs.
New features to help ensure your apps are performant and reliable, include DC/OS-native tools for monitoring and troubleshooting.Extending our capabilities within container orchestration, we've broadened workload support with GPU-based scheduling and Pods – which allows you to easily launch and manage sets of both traditional linux and docker containers within DC/OS. Lastly, security and compliance is a big piece of Mesosphere DC/OS 1.9, which enables operators to give developers access to debugging and associated apps while limiting access to the underlying infrastructure. Mesosphere DC/OS 1.9 has also been independently audited for compliance against NIST, ISO, FISMA, OSC and HIPPA certifications. Now in early access, Mesosphere DC/OS 1.9 will be GA before the end of this month.
Mesosphere DC/OS technologies already power more containers in production than any other open source-based software in the world, and 1.9 further reinforces our enterprise-grade solution for modern, data-rich applications. In fact, in the Mesos and DC/OS communities, 62% already run containers in production as of August 2016.
Read on below for more details about these new features.
Run Any of the Data Services You Need on One Single Platform
Instead of having to learn and implement the unique deployment and scalability requirements of distributed databases or big data services, Mesosphere DC/OS lets you simply click and deploy any of these applications in seconds - not the weeks required with manual traditional methods. As of version 1.9, new data services can now be quickly deployed and run elastically on Mesosphere DC/OS, accessible from the DC/OS Universe service catalog, which now includes over 100 services. The new data services are from DataStax (company behind Apache Cassandra), Couchbase, Redis Labs, Elastic, and Alluxio, who have build their DC/OS service using our Partner SDK. This is on top of data services already in the catalog including Confluent Platform, Apache Cassandra, Kafka, Spark, and HDFS.
In addition, as these databases are deployed natively as protected services, you reduce the risks of losing data, and automation ensures you can scale and upgrade without service disruption. Mesosphere DC/OS makes deploying and using data services as easy those you find from public cloud providers, but with the power to deploy them and store your data on any infrastructure you choose.
Monitoring and Troubleshooting for Distributed Applications
Finding, aggregating, and analyzing metrics for modern distributed apps is incredibly difficult, as critical information like logs and metrics come from several places such as application containers, container orchestrators, host operating systems, and more. Mesosphere DC/OS 1.9 simplifies the collection of logs and metrics, and enriches this information with user and service data to provide context. By making this aggregated data easily pluggable with your existing monitoring tools (e.g., Splunk, ELK, Datadog, Prometheus, or custom-developed monitoring tools), with DC/OS 1.9 you can now get a comprehensive view into the health of your application and cluster. Additionally, DC/OS can tell you why apps failed to run (e.g., port issue, no capacity).
In addition, modern applications are dynamically distributed across containers, so it can be hard to troubleshoot and isolate problems when they occur. Some issues require a live connection to debug, and developers often don't have direct access to server hosts. Mesosphere DC/OS 1.9 introduces container process injection, which allows developers to remotely run commands in any container (e.g., ping, bash, java debugger) in the same namespace as the task being investigated.
Broad Workload Support: From Traditional Apps to Machine Learning
Deploying traditional N-tier applications on containers can be difficult, making teams lose out on the benefits of containerization with legacy apps. The addition of Pods within DC/OS now makes it easier than ever to scale these applications and ensure maximum uptime. With pods, the web, app, and DB tiers can share hosts while using dedicated IPs, always running on hosts attached to their storage volumes. This allows you to manage your traditional apps within DC/OS to easily scale to any capacity. Dynamic load balancing & service discovery built natively within the platform ensures app services are always connected, and automatic failure detection and recovery ensures apps are always running.
Machine learning is the next evolution of data-driven applications and involves training neural networks, which can take hours or weeks. With GPU acceleration, neural net training can be 10-20 faster than with CPUs. DC/OS now allows you to isolate and reserve GPU resources for the workloads that need them, but also pool these GPU-enabled resources with the rest of the infrastructure for higher overall utilization.
Secure and Compliant Operations
Security and compliance is a key concern for enterprises, and Mesosphere DC/OS 1.9 delivers. In addition to existing authentication, authorization and multi-tenancy capabilities for applications and services, DC/OS 1.9 integrates with multiple identity providers. Administrators can also determine which users have access to metrics, logging and debugging tools for applications (using access control lists) while protecting the underlying infrastructure. On compliance, Mesosphere DC/OS 1.9 has been independently audited for compliance with ISO-27001, SOC-2, NIST-800/53, FISMA-Mod., and HIPAA standards.
For more details on the DC/OS 1.9 release, please read the release notes.
