'Data center OS' Mesophere plans 2015 hire and roadmap drive
Integrated dream for big-data brains
When Mesosphere landed $36m VC funding in December 2014 it also released the first part of its Datacenter Operating System (DCOS).
Mesosphere is now in the middle of a recruitment drive, looking to hit a headcount of between 100 and 150 by the end of 2015 – up from 52.
Engineering and product development are the other priority for that cash and this quarter Mesosphere reckons on delivering a new version of DCOS – it’s just not clear on what the nomenclature will be.
Company co-founder Florian Leibert speaking to The Reg last month categorized what had shipped so far was pieces you had to assemble and that Mesosphere custom engineered.
The goal is for a full package system with command line-interface, native Mesos apps, user interface, and integration with some clouds.
The start-up has been working with Google Compute among others to date.
The idea is to make it more convenient to install DCOS on your own data center or in the cloud.
The bigger goal for 2015 is to integrate DCOS with applications such as the Apache Cassandra NoSQL database and Apache’s Kafka for streaming analytics.
“There’s a lot of work to be done,” Leibert said. “Last year was integration with containers and the creation of some primitives."
"We want to make this a compelling solution to anyone who wants to launch a website or do analytics from their cluster."
So far in 2015, DCOS has been connected to Apache’s YARN through a project called Myriad build with MapR Technologies and eBay. YARN is the cluster management technology used in the latest release of Hadoop.
The principle behind Mesos is to consolidate big-data workloads on a single pool of servers and resources, without the need to spin up separate servers for different jobs. This simplifies the building and running of software and services.
Also, the idea is for a single programming framework for development, so devs aren't building to different server and data center resources.
Myriad lets you run YARN jobs alongside other workloads on the same Mesos cluster. You can also run more than one instance of YARN on the same cluster, in parallel, again, without need for separate large clusters.
The plan at time of speaking was to submit Myriad to Apache as an incubator project during the first quarter, although that doesn’t seem to have happened yet. ®