This allows applications, which requires different service runtimes with their relationship and dependencies, to be deployed. Furthermore, each of the service runtimes in the application can scale independently or jointly with the dependent services. The following operations are supported in composite application support:
Starting up instances using the StartupOrder, which is defined in the cartridge group definition.
Starting up instances using the StartupOrder, which is defined in the application definition. After the relevant clusters and groups get activated according to the startup order, the application itself gets activated.
Termination based on the TerminationBehaviour, which is defined in the cartridge group definition.
Termination based on the TerminationBehaviour, which is defined in the application definition.
Sharing information between instances when one instance is dependent on another.
Docker support using Google Kubernetes and CoreOS. Thereby, Stratos can also leverage the use of Docker in a PaaS. The following aspects are supported for Docker:
Auto-scaling Docker Containers.
Manual scaling Docker Containers.
Git based artifact deployment for Docker.
CLI support for Docker deployments.
VM LB support for dockers
Private Docker registry
A Python based and Java based cartridge agent is available in Stratos.
Update artifacts at run-time
After an application is deployed, Stratos allows users to update the following artifacts, which directly effects the runtime.
Auto-scaling policy definition
Deployment policy definition
Cloud Controller (CC) leverages Apache jclouds' APIs and provides a generic interface to communicate with different IaaSes.
The Auto-scaler uses a Complex Event Processor (CEP) for real-time decision making, and it integrates both real-time and rule-base decision making to provide better control over scaling of platforms. Stratos allows users to define auto-scaling policies with multiple factors,i.e., requests in flight, memory consumption and load average, which are considered when scaling up or down. The Auto-scaler also supports scaling for non-HTTP transport.
Multiple IaaS support
Apache Stratos is tested on the following IaaS providers: AWS EC2 and OpenStack. However, it is very easy to extend Apache Stratos to support any IaaS that is supported by Apache jclouds i.e., Google cloud, CloudStack etc.
Apache Stratos supports multiple IaaSs. When the maximum limit of instances have been reached in an IaaS, instances are spawned on another IaaS, which is in another network partition. Thereby, this will enable resource peak times to be off-loaded to another cloud.
Controlling IaaS resources
It is possible for DevOps to define partitions in a network partition, to control IaaS resources. Thereby, Apache Stratos can control resources per cloud, region, and zone. Controlling of IaaS resources provide a high availability and solves disaster recovery concerns. For more information, see [Cloud Partitioning] (https://cwiki.apache.org/confluence/display/STRATOS/4.1.x+Cloud+Partitioning).
Loosely coupled communication
Stratos uses the Advanced Message Queuing Protocol (AMQP) messaging technology for communication among all its components. Apache Stratos uses an AMQP Message Broker (MB), namely ActiveMQ, to communicate in a loosely coupled fashion. However, it is possible to use any MB, which supports AMQP, with Stratos.
Stratos supports in-container multi-tenancy. Thereby, this helps to optimize the resource utilization.
Support for PHP, MySQL, Tomcat, Windows based (.NET) cartridges. The following is the list of tested cartridges: PHP, MySQL, Ruby, Node.js, Wordpress, Drupal, Tomcat, HAProxy and NGINX.
Pluggable architecture support for cartridges
A cartridge is a package of code that includes a Virtual Machine (VM) image plus additional configuration, which can be plugged into Stratos to offer a new PaaS service. Stratos supports single tenant and multi-tenant cartridges. If needed, tenants can easily add their own cartridges to Stratos.
Cartridge automation using Puppet
Cartridges can be easily configured with the use of an orchestration layer such as Puppet.
Support for third party load balancers
Stratos supports third-party load balancers (LBs), i.e, HAProxy, NGINX. Thereby, if required, users can use their own LB with Stratos.
Monitoring and metering
Apache Stratos provides centralized monitoring and metering. The level of resource utilization in Stratos is measured using metering.
Persistent volume support for cartridges
If required, the DevOps can enable a persistent volume for cartridges. If persistent volume is enabled, Apache Stratos automatically attaches a volume when a new cartridge instance is created.
Gracefully shutdown instances
Before terminating an instance, when scaling down, the Auto-scaler will allow all the existing requests to the instance to gracefully shutdown, and not accepting any new requests for that instance.
MQTT and AMQP support
Copyright (C) 2013-2015 The Apache Software Foundation
Licensed under the Apache License, Version 2.0