Here we are, on the cusp of that crazy “last week before EMC World”. If you’ve ever working inside a company that does these mega-events (think VMworld, Oracle Open World, etc) – you know what I’m talking about.
If you haven’t – let me summarize: insanity. You have major product launches(and we have some doozies) lining up and product/engineering teams working overtime; you have marketing/communications teams working like crazy; demos being created/destroyed/rebuilt; hands-on lab environments being pushed to the limits; and of course you have schedules going nuts with customers and partners.
And yes you have last-minute keynote madness too – with stunts, pyrotechnics, and “kapow” moments galore people are working on.
There are constant moments that are like the “Martian” or “Apollo 13” - far less life/death of course, but just as creative “let’s work around the problem”).
There’s something in the air this time that’s a little different. Hard to put my finger on, but I think it’s an extra little bit of energy because in all likelyhood, this could be the last standalone EMC World, as the Dell/EMC merger moves closer and virtually certain.
Remember – register for EMC World here – you won’t regret it!
- ScaleIO is awesome. It is the industries best (IMO) transactional SDS which is hypervisor/OS neutral (smoking the performance/cost/complexity of other transactional SDS stacks). ScaleIO is a GREAT choice for customers looking for what storage target they want to use with Cinder and Nova-Compute support (of course we have broad Cinder support for all EMC platforms like VNX, VMAX, XtremIO if customers also want to use what they already have).
- ScaleIO is available in 3 forms:
- Software-only ScaleIO – bring your own HW – which you can get freely and frictionlessly here
- Bundled with tested/supported hardware and an OS – these are VxRack Flex Nodes.
- Fully-engineered Rack-Scale HCI form – these are VxRack System 1000 with Flex Nodes.
- That last one – VxRack System 1000 with Flex Nodes includes everything (ToR, management network, Flex Nodes using ScaleIO, VxRack Manager – which is the management and orchestration software, and VCE Vision software) – all engineered, designed, supported, and sustained as one.*
*Note that VxRack System 1000 is not a singular “thing”. It is a Rack-Scale system design, and has several node variations depending on use cases. Every one is defined by being a system designed to scale at Rack-Scale, and like all engineered converged systems, contemplates the networking (physical and SDN) domain as part of the system design. The Node types are selected based on use case. Flex Nodes are targeted at cases where customers want the flexibility to have workloads that use vSphere, KVM, Physical Nodes, and in the near future Hyper-V, and hence need an open SDS – ScaleIO. SDDC Nodes are targeted for customer uniquely want vSphere-based abstraction, and hence use the SDS which is most integrated with vSphere – VSAN. VxRack System 1000 with SDDC nodes has been quotable for about a month, and ship at the end of May. Hint: the VxRack System 1000 family will expand at EMC World – exciting news to come!
What I think is cool is to see how ScaleIO is deepening integration with the OpenStack ecosystem – with Mirantis Fuel plugins, and now with Canonical Juju charms. If you come to the OpenStack summit, we’ll be sharing more what we’re doing down this front.
This is all still in Beta – but I think part of the fun here is that the software is accessible, the repos are open, and some readers want to play/contribute.
What’s new in Beta 2 (thanks Alexandre Levine for the great work so far):
- Management and internal interfaces for MDM cluster
- Storage and management interfaces for SDSs
- Thin and thick volumes support
- JuJu scaleio-openstack now explicitly connects to both cinder and nova-compute, patched nova-compute for this is put into our charms repo.
- nova-compute and cinder configurations changes now survive reboot and other config changes
- apt-repo and ftp can be specified in configuration
- Haproxy is supported in ScaleIO Gateway and in Fuel
- Performance tweaks were introduced for ScaleIO cluster
The suite provides integration with Canonical JuJu and Mirantis Fuel for deployment of ScaleIO 2.0 with OpenStack.
It allows provisioning and further configuration and maintenance according to capabilities providing by particular deployment technology (Fuel and JuJu).
The following repositories contain the artefacts:
- JuJu charms: source code: https://github.com/cloudscaling/juju-scaleio
- Fuel puppet plugin: source code: https://github.com/cloudscaling/fuel-plugin-scaleio
- OpenStack integration puppet: source code: https://github.com/cloudscaling/puppet-scaleio & puppetforge: https://forge.puppetlabs.com/cloudscaling/scaleio
- ScaleIO deployment puppet: source code: https://github.com/cloudscaling/puppet-scaleio & puppetforge: https://forge.puppetlabs.com/cloudscaling/scaleio_openstack
- BTW – not expressly related, but if you want Ansible for ScaleIO – get it here: https://github.com/sperreault/ansible-scaleio; and you can get get Go API bindings here: https://github.com/emccode/goscaleio
- JuJu deployment requires JuJu, JuJu charms, OpenStack integration puppet and ScaleIO deployment puppet.
- Fuel deployment requires Mirantis Fuel, Fuel puppet plugin, OpenStack integration puppet and ScaleIO deployment puppet.
The features provided for both:
- Full support of existing ScaleIO to OpenStack integration for volumes in cinder and nova
- Ephemeral storage backed instances support for nova.
- Support for resize and live migration for such instances.
OpenStack versions supported
- Juno, nova:
- Kilo, nova:
- Liberty, nova:
Mirantis Fuel versions supported
Fuel 6.1. 7.0
Operating Systems supported
Again – thanks Alexandre – and have fun all!
… Now for those of you who love this kind of idea but don’t see the fun in playing/building and/or don’t have all the time/resources to play (and just want to get on to the business of getting developers going faster – we have something interesting in store for you – STAY TUNED!